Merge remote-tracking branch 'origin/main'

wucong.lyb

2023-07-04 6b840d51437d3b23854bd9bed84cfb0d8976d079

Merge remote-tracking branch 'origin/main'

# Conflicts:
#	funasr/runtime/run_server.sh

 README.md

@@ -20,9 +20,18 @@
| [**M2MET2.0 Challenge**](https://github.com/alibaba-damo-academy/FunASR#multi-channel-multi-party-meeting-transcription-20-m2met20-challenge)

## What's new: 

### FunASR runtime-SDK

- 2023.07.03: 
We have release the FunASR runtime-SDK-0.1.0, file transcription service (Mandarin) is now supported ([ZH](funasr/runtime/readme_cn.md)/[EN](funasr/runtime/readme.md))

### Multi-Channel Multi-Party Meeting Transcription 2.0 (M2MeT2.0) Challenge

We are pleased to announce that the M2MeT2.0 challenge has been accepted by the ASRU 2023 challenge special session. The registration is now open. The baseline system is conducted on FunASR and is provided as a receipe of AliMeeting corpus. For more details you can see the guidence of M2MET2.0 ([CN](https://alibaba-damo-academy.github.io/FunASR/m2met2_cn/index.html)/[EN](https://alibaba-damo-academy.github.io/FunASR/m2met2/index.html)).

### Release notes

For the release notes, please ref to [news](https://github.com/alibaba-damo-academy/FunASR/releases)

## Highlights

 docs/images/dingding.jpg

 funasr/bin/asr_inference_launch.py

@@ -1604,6 +1604,8 @@
        return inference_mfcca(**kwargs)
    elif mode == "rnnt":
        return inference_transducer(**kwargs)
    elif mode == "bat":
        return inference_transducer(**kwargs)
    elif mode == "sa_asr":
        return inference_sa_asr(**kwargs)
    else:

 funasr/build_utils/build_asr_model.py

@@ -26,6 +26,7 @@
from funasr.models.e2e_asr_mfcca import MFCCA

from funasr.models.e2e_asr_transducer import TransducerModel, UnifiedTransducerModel
from funasr.models.e2e_asr_bat import BATModel

from funasr.models.e2e_sa_asr import SAASRModel
from funasr.models.e2e_asr_paraformer import Paraformer, ParaformerOnline, ParaformerBert, BiCifParaformer, ContextualParaformer
@@ -46,7 +47,7 @@
from funasr.models.frontend.wav_frontend import WavFrontend
from funasr.models.frontend.windowing import SlidingWindow
from funasr.models.joint_net.joint_network import JointNetwork
from funasr.models.predictor.cif import CifPredictor, CifPredictorV2, CifPredictorV3
from funasr.models.predictor.cif import CifPredictor, CifPredictorV2, CifPredictorV3, BATPredictor
from funasr.models.specaug.specaug import SpecAug
from funasr.models.specaug.specaug import SpecAugLFR
from funasr.modules.subsampling import Conv1dSubsampling
@@ -99,7 +100,7 @@
        rnnt=TransducerModel,
        rnnt_unified=UnifiedTransducerModel,
        sa_asr=SAASRModel,

        bat=BATModel,
    ),
    default="asr",
)
@@ -188,6 +189,7 @@
        ctc_predictor=None,
        cif_predictor_v2=CifPredictorV2,
        cif_predictor_v3=CifPredictorV3,
        bat_predictor=BATPredictor,
    ),
    default="cif_predictor",
    optional=True,
@@ -313,12 +315,15 @@
    encoder = encoder_class(input_size=input_size, **args.encoder_conf)

    # decoder
    decoder_class = decoder_choices.get_class(args.decoder)
    decoder = decoder_class(
        vocab_size=vocab_size,
        encoder_output_size=encoder.output_size(),
        **args.decoder_conf,
    )
    if hasattr(args, "decoder") and args.decoder is not None:
        decoder_class = decoder_choices.get_class(args.decoder)
        decoder = decoder_class(
            vocab_size=vocab_size,
            encoder_output_size=encoder.output_size(),
            **args.decoder_conf,
        )
    else:
        decoder = None

    # ctc
    ctc = CTC(
@@ -463,6 +468,53 @@
            joint_network=joint_network,
            **args.model_conf,
        )
    elif args.model == "bat":
        # 5. Decoder
        encoder_output_size = encoder.output_size()

        rnnt_decoder_class = rnnt_decoder_choices.get_class(args.rnnt_decoder)
        decoder = rnnt_decoder_class(
            vocab_size,
            **args.rnnt_decoder_conf,
        )
        decoder_output_size = decoder.output_size

        if getattr(args, "decoder", None) is not None:
            att_decoder_class = decoder_choices.get_class(args.decoder)

            att_decoder = att_decoder_class(
                vocab_size=vocab_size,
                encoder_output_size=encoder_output_size,
                **args.decoder_conf,
            )
        else:
            att_decoder = None
        # 6. Joint Network
        joint_network = JointNetwork(
            vocab_size,
            encoder_output_size,
            decoder_output_size,
            **args.joint_network_conf,
        )

        predictor_class = predictor_choices.get_class(args.predictor)
        predictor = predictor_class(**args.predictor_conf)

        model_class = model_choices.get_class(args.model)
        # 7. Build model
        model = model_class(
            vocab_size=vocab_size,
            token_list=token_list,
            frontend=frontend,
            specaug=specaug,
            normalize=normalize,
            encoder=encoder,
            decoder=decoder,
            att_decoder=att_decoder,
            joint_network=joint_network,
            predictor=predictor,
            **args.model_conf,
        )
    elif args.model == "sa_asr":
        asr_encoder_class = asr_encoder_choices.get_class(args.asr_encoder)
        asr_encoder = asr_encoder_class(input_size=input_size, **args.asr_encoder_conf)

 funasr/models/e2e_asr_bat.py

New file
@@ -0,0 +1,496 @@
"""Boundary Aware Transducer (BAT) model."""

import logging
from contextlib import contextmanager
from typing import Dict, List, Optional, Tuple, Union

import torch
from packaging.version import parse as V
from funasr.losses.label_smoothing_loss import (
    LabelSmoothingLoss,  # noqa: H301
)
from funasr.models.frontend.abs_frontend import AbsFrontend
from funasr.models.specaug.abs_specaug import AbsSpecAug
from funasr.models.decoder.rnnt_decoder import RNNTDecoder
from funasr.models.decoder.abs_decoder import AbsDecoder as AbsAttDecoder
from funasr.models.encoder.abs_encoder import AbsEncoder
from funasr.models.joint_net.joint_network import JointNetwork
from funasr.modules.nets_utils import get_transducer_task_io
from funasr.modules.nets_utils import th_accuracy
from funasr.modules.nets_utils import make_pad_mask
from funasr.modules.add_sos_eos import add_sos_eos
from funasr.layers.abs_normalize import AbsNormalize
from funasr.torch_utils.device_funcs import force_gatherable
from funasr.models.base_model import FunASRModel

if V(torch.__version__) >= V("1.6.0"):
    from torch.cuda.amp import autocast
else:

    @contextmanager
    def autocast(enabled=True):
        yield


class BATModel(FunASRModel):
    """BATModel module definition.

    Args:
        vocab_size: Size of complete vocabulary (w/ EOS and blank included).
        token_list: List of token
        frontend: Frontend module.
        specaug: SpecAugment module.
        normalize: Normalization module.
        encoder: Encoder module.
        decoder: Decoder module.
        joint_network: Joint Network module.
        transducer_weight: Weight of the Transducer loss.
        fastemit_lambda: FastEmit lambda value.
        auxiliary_ctc_weight: Weight of auxiliary CTC loss.
        auxiliary_ctc_dropout_rate: Dropout rate for auxiliary CTC loss inputs.
        auxiliary_lm_loss_weight: Weight of auxiliary LM loss.
        auxiliary_lm_loss_smoothing: Smoothing rate for LM loss' label smoothing.
        ignore_id: Initial padding ID.
        sym_space: Space symbol.
        sym_blank: Blank Symbol
        report_cer: Whether to report Character Error Rate during validation.
        report_wer: Whether to report Word Error Rate during validation.
        extract_feats_in_collect_stats: Whether to use extract_feats stats collection.

    """

    def __init__(
        self,
        vocab_size: int,
        token_list: Union[Tuple[str, ...], List[str]],
        frontend: Optional[AbsFrontend],
        specaug: Optional[AbsSpecAug],
        normalize: Optional[AbsNormalize],
        encoder: AbsEncoder,
        decoder: RNNTDecoder,
        joint_network: JointNetwork,
        att_decoder: Optional[AbsAttDecoder] = None,
        predictor = None,
        transducer_weight: float = 1.0,
        predictor_weight: float = 1.0,
        cif_weight: float = 1.0,
        fastemit_lambda: float = 0.0,
        auxiliary_ctc_weight: float = 0.0,
        auxiliary_ctc_dropout_rate: float = 0.0,
        auxiliary_lm_loss_weight: float = 0.0,
        auxiliary_lm_loss_smoothing: float = 0.0,
        ignore_id: int = -1,
        sym_space: str = "<space>",
        sym_blank: str = "<blank>",
        report_cer: bool = True,
        report_wer: bool = True,
        extract_feats_in_collect_stats: bool = True,
        lsm_weight: float = 0.0,
        length_normalized_loss: bool = False,
        r_d: int = 5,
        r_u: int = 5,
    ) -> None:
        """Construct an BATModel object."""
        super().__init__()

        # The following labels ID are reserved: 0 (blank) and vocab_size - 1 (sos/eos)
        self.blank_id = 0
        self.vocab_size = vocab_size
        self.ignore_id = ignore_id
        self.token_list = token_list.copy()

        self.sym_space = sym_space
        self.sym_blank = sym_blank

        self.frontend = frontend
        self.specaug = specaug
        self.normalize = normalize

        self.encoder = encoder
        self.decoder = decoder
        self.joint_network = joint_network

        self.criterion_transducer = None
        self.error_calculator = None

        self.use_auxiliary_ctc = auxiliary_ctc_weight > 0
        self.use_auxiliary_lm_loss = auxiliary_lm_loss_weight > 0

        if self.use_auxiliary_ctc:
            self.ctc_lin = torch.nn.Linear(encoder.output_size(), vocab_size)
            self.ctc_dropout_rate = auxiliary_ctc_dropout_rate

        if self.use_auxiliary_lm_loss:
            self.lm_lin = torch.nn.Linear(decoder.output_size, vocab_size)
            self.lm_loss_smoothing = auxiliary_lm_loss_smoothing

        self.transducer_weight = transducer_weight
        self.fastemit_lambda = fastemit_lambda

        self.auxiliary_ctc_weight = auxiliary_ctc_weight
        self.auxiliary_lm_loss_weight = auxiliary_lm_loss_weight

        self.report_cer = report_cer
        self.report_wer = report_wer

        self.extract_feats_in_collect_stats = extract_feats_in_collect_stats

        self.criterion_pre = torch.nn.L1Loss()
        self.predictor_weight = predictor_weight
        self.predictor = predictor
        
        self.cif_weight = cif_weight
        if self.cif_weight > 0:
            self.cif_output_layer = torch.nn.Linear(encoder.output_size(), vocab_size)
            self.criterion_cif = LabelSmoothingLoss(
                size=vocab_size,
                padding_idx=ignore_id,
                smoothing=lsm_weight,
                normalize_length=length_normalized_loss,
            )        
        self.r_d = r_d
        self.r_u = r_u

    def forward(
        self,
        speech: torch.Tensor,
        speech_lengths: torch.Tensor,
        text: torch.Tensor,
        text_lengths: torch.Tensor,
        **kwargs,
    ) -> Tuple[torch.Tensor, Dict[str, torch.Tensor], torch.Tensor]:
        """Forward architecture and compute loss(es).

        Args:
            speech: Speech sequences. (B, S)
            speech_lengths: Speech sequences lengths. (B,)
            text: Label ID sequences. (B, L)
            text_lengths: Label ID sequences lengths. (B,)
            kwargs: Contains "utts_id".

        Return:
            loss: Main loss value.
            stats: Task statistics.
            weight: Task weights.

        """
        assert text_lengths.dim() == 1, text_lengths.shape
        assert (
            speech.shape[0]
            == speech_lengths.shape[0]
            == text.shape[0]
            == text_lengths.shape[0]
        ), (speech.shape, speech_lengths.shape, text.shape, text_lengths.shape)

        batch_size = speech.shape[0]
        text = text[:, : text_lengths.max()]

        # 1. Encoder
        encoder_out, encoder_out_lens = self.encode(speech, speech_lengths)
        if hasattr(self.encoder, 'overlap_chunk_cls') and self.encoder.overlap_chunk_cls is not None:
            encoder_out, encoder_out_lens = self.encoder.overlap_chunk_cls.remove_chunk(encoder_out, encoder_out_lens,
                                                                                        chunk_outs=None)

        encoder_out_mask = (~make_pad_mask(encoder_out_lens, maxlen=encoder_out.size(1))[:, None, :]).to(encoder_out.device)
        # 2. Transducer-related I/O preparation
        decoder_in, target, t_len, u_len = get_transducer_task_io(
            text,
            encoder_out_lens,
            ignore_id=self.ignore_id,
        )

        # 3. Decoder
        self.decoder.set_device(encoder_out.device)
        decoder_out = self.decoder(decoder_in, u_len)

        pre_acoustic_embeds, pre_token_length, _, pre_peak_index = self.predictor(encoder_out, text, encoder_out_mask, ignore_id=self.ignore_id)
        loss_pre = self.criterion_pre(text_lengths.type_as(pre_token_length), pre_token_length)

        if self.cif_weight > 0.0:
            cif_predict = self.cif_output_layer(pre_acoustic_embeds)
            loss_cif = self.criterion_cif(cif_predict, text)
        else:
            loss_cif = 0.0

        # 5. Losses
        boundary = torch.zeros((encoder_out.size(0), 4), dtype=torch.int64, device=encoder_out.device)
        boundary[:, 2] = u_len.long().detach()
        boundary[:, 3] = t_len.long().detach()

        pre_peak_index = torch.floor(pre_peak_index).long()
        s_begin = pre_peak_index - self.r_d

        T = encoder_out.size(1)
        B = encoder_out.size(0)
        U = decoder_out.size(1)

        mask = torch.arange(0, T, device=encoder_out.device).reshape(1, T).expand(B, T)
        mask = mask <= boundary[:, 3].reshape(B, 1) - 1

        s_begin_padding = boundary[:, 2].reshape(B, 1) - (self.r_u+self.r_d) + 1
        # handle the cases where `len(symbols) < s_range`
        s_begin_padding = torch.clamp(s_begin_padding, min=0)

        s_begin = torch.where(mask, s_begin, s_begin_padding)
        
        mask2 = s_begin <  boundary[:, 2].reshape(B, 1) - (self.r_u+self.r_d) + 1

        s_begin = torch.where(mask2, s_begin, boundary[:, 2].reshape(B, 1) - (self.r_u+self.r_d) + 1)

        s_begin = torch.clamp(s_begin, min=0)
        
        ranges = s_begin.reshape((B, T, 1)).expand((B, T, min(self.r_u+self.r_d, min(u_len)))) + torch.arange(min(self.r_d+self.r_u, min(u_len)), device=encoder_out.device)

        import fast_rnnt
        am_pruned, lm_pruned = fast_rnnt.do_rnnt_pruning(
            am=self.joint_network.lin_enc(encoder_out),
            lm=self.joint_network.lin_dec(decoder_out),
            ranges=ranges,
        )

        logits = self.joint_network(am_pruned, lm_pruned, project_input=False)

        with torch.cuda.amp.autocast(enabled=False):
            loss_trans = fast_rnnt.rnnt_loss_pruned(
                logits=logits.float(),
                symbols=target.long(),
                ranges=ranges,
                termination_symbol=self.blank_id,
                boundary=boundary,
                reduction="sum",
            )

        cer_trans, wer_trans = None, None
        if not self.training and (self.report_cer or self.report_wer):
            if self.error_calculator is None:
                from funasr.modules.e2e_asr_common import ErrorCalculatorTransducer as ErrorCalculator
                self.error_calculator = ErrorCalculator(
                    self.decoder,
                    self.joint_network,
                    self.token_list,
                    self.sym_space,
                    self.sym_blank,
                    report_cer=self.report_cer,
                    report_wer=self.report_wer,
                )
            cer_trans, wer_trans = self.error_calculator(encoder_out, target, t_len)

        loss_ctc, loss_lm = 0.0, 0.0

        if self.use_auxiliary_ctc:
            loss_ctc = self._calc_ctc_loss(
                encoder_out,
                target,
                t_len,
                u_len,
            )

        if self.use_auxiliary_lm_loss:
            loss_lm = self._calc_lm_loss(decoder_out, target)

        loss = (
            self.transducer_weight * loss_trans
            + self.auxiliary_ctc_weight * loss_ctc
            + self.auxiliary_lm_loss_weight * loss_lm
            + self.predictor_weight * loss_pre
            + self.cif_weight * loss_cif
        )

        stats = dict(
            loss=loss.detach(),
            loss_transducer=loss_trans.detach(),
            loss_pre=loss_pre.detach(),
            loss_cif=loss_cif.detach() if loss_cif > 0.0 else None,
            aux_ctc_loss=loss_ctc.detach() if loss_ctc > 0.0 else None,
            aux_lm_loss=loss_lm.detach() if loss_lm > 0.0 else None,
            cer_transducer=cer_trans,
            wer_transducer=wer_trans,
        )

        # force_gatherable: to-device and to-tensor if scalar for DataParallel
        loss, stats, weight = force_gatherable((loss, stats, batch_size), loss.device)

        return loss, stats, weight

    def collect_feats(
        self,
        speech: torch.Tensor,
        speech_lengths: torch.Tensor,
        text: torch.Tensor,
        text_lengths: torch.Tensor,
        **kwargs,
    ) -> Dict[str, torch.Tensor]:
        """Collect features sequences and features lengths sequences.

        Args:
            speech: Speech sequences. (B, S)
            speech_lengths: Speech sequences lengths. (B,)
            text: Label ID sequences. (B, L)
            text_lengths: Label ID sequences lengths. (B,)
            kwargs: Contains "utts_id".

        Return:
            {}: "feats": Features sequences. (B, T, D_feats),
                "feats_lengths": Features sequences lengths. (B,)

        """
        if self.extract_feats_in_collect_stats:
            feats, feats_lengths = self._extract_feats(speech, speech_lengths)
        else:
            # Generate dummy stats if extract_feats_in_collect_stats is False
            logging.warning(
                "Generating dummy stats for feats and feats_lengths, "
                "because encoder_conf.extract_feats_in_collect_stats is "
                f"{self.extract_feats_in_collect_stats}"
            )

            feats, feats_lengths = speech, speech_lengths

        return {"feats": feats, "feats_lengths": feats_lengths}

    def encode(
        self,
        speech: torch.Tensor,
        speech_lengths: torch.Tensor,
    ) -> Tuple[torch.Tensor, torch.Tensor]:
        """Encoder speech sequences.

        Args:
            speech: Speech sequences. (B, S)
            speech_lengths: Speech sequences lengths. (B,)

        Return:
            encoder_out: Encoder outputs. (B, T, D_enc)
            encoder_out_lens: Encoder outputs lengths. (B,)

        """
        with autocast(False):
            # 1. Extract feats
            feats, feats_lengths = self._extract_feats(speech, speech_lengths)

            # 2. Data augmentation
            if self.specaug is not None and self.training:
                feats, feats_lengths = self.specaug(feats, feats_lengths)

            # 3. Normalization for feature: e.g. Global-CMVN, Utterance-CMVN
            if self.normalize is not None:
                feats, feats_lengths = self.normalize(feats, feats_lengths)

        # 4. Forward encoder
        encoder_out, encoder_out_lens, _ = self.encoder(feats, feats_lengths)

        assert encoder_out.size(0) == speech.size(0), (
            encoder_out.size(),
            speech.size(0),
        )
        assert encoder_out.size(1) <= encoder_out_lens.max(), (
            encoder_out.size(),
            encoder_out_lens.max(),
        )

        return encoder_out, encoder_out_lens

    def _extract_feats(
        self, speech: torch.Tensor, speech_lengths: torch.Tensor
    ) -> Tuple[torch.Tensor, torch.Tensor]:
        """Extract features sequences and features sequences lengths.

        Args:
            speech: Speech sequences. (B, S)
            speech_lengths: Speech sequences lengths. (B,)

        Return:
            feats: Features sequences. (B, T, D_feats)
            feats_lengths: Features sequences lengths. (B,)

        """
        assert speech_lengths.dim() == 1, speech_lengths.shape

        # for data-parallel
        speech = speech[:, : speech_lengths.max()]

        if self.frontend is not None:
            feats, feats_lengths = self.frontend(speech, speech_lengths)
        else:
            feats, feats_lengths = speech, speech_lengths

        return feats, feats_lengths

    def _calc_ctc_loss(
        self,
        encoder_out: torch.Tensor,
        target: torch.Tensor,
        t_len: torch.Tensor,
        u_len: torch.Tensor,
    ) -> torch.Tensor:
        """Compute CTC loss.

        Args:
            encoder_out: Encoder output sequences. (B, T, D_enc)
            target: Target label ID sequences. (B, L)
            t_len: Encoder output sequences lengths. (B,)
            u_len: Target label ID sequences lengths. (B,)

        Return:
            loss_ctc: CTC loss value.

        """
        ctc_in = self.ctc_lin(
            torch.nn.functional.dropout(encoder_out, p=self.ctc_dropout_rate)
        )
        ctc_in = torch.log_softmax(ctc_in.transpose(0, 1), dim=-1)

        target_mask = target != 0
        ctc_target = target[target_mask].cpu()

        with torch.backends.cudnn.flags(deterministic=True):
            loss_ctc = torch.nn.functional.ctc_loss(
                ctc_in,
                ctc_target,
                t_len,
                u_len,
                zero_infinity=True,
                reduction="sum",
            )
        loss_ctc /= target.size(0)

        return loss_ctc

    def _calc_lm_loss(
        self,
        decoder_out: torch.Tensor,
        target: torch.Tensor,
    ) -> torch.Tensor:
        """Compute LM loss.

        Args:
            decoder_out: Decoder output sequences. (B, U, D_dec)
            target: Target label ID sequences. (B, L)

        Return:
            loss_lm: LM loss value.

        """
        lm_loss_in = self.lm_lin(decoder_out[:, :-1, :]).view(-1, self.vocab_size)
        lm_target = target.view(-1).type(torch.int64)

        with torch.no_grad():
            true_dist = lm_loss_in.clone()
            true_dist.fill_(self.lm_loss_smoothing / (self.vocab_size - 1))

            # Ignore blank ID (0)
            ignore = lm_target == 0
            lm_target = lm_target.masked_fill(ignore, 0)

            true_dist.scatter_(1, lm_target.unsqueeze(1), (1 - self.lm_loss_smoothing))

        loss_lm = torch.nn.functional.kl_div(
            torch.log_softmax(lm_loss_in, dim=1),
            true_dist,
            reduction="none",
        )
        loss_lm = loss_lm.masked_fill(ignore.unsqueeze(1), 0).sum() / decoder_out.size(
            0
        )

        return loss_lm

 funasr/models/e2e_asr_transducer.py

@@ -353,11 +353,6 @@
        """
        if self.criterion_transducer is None:
            try:
                # from warprnnt_pytorch import RNNTLoss
            # self.criterion_transducer = RNNTLoss(
                    # reduction="mean",
                    # fastemit_lambda=self.fastemit_lambda,
                # )
                from warp_rnnt import rnnt_loss as RNNTLoss
                self.criterion_transducer = RNNTLoss

@@ -368,12 +363,6 @@
                )
                exit(1)

        # loss_transducer = self.criterion_transducer(
        #     joint_out,
        #     target,
        #     t_len,
        #     u_len,
        # )
        log_probs = torch.log_softmax(joint_out, dim=-1)

        loss_transducer = self.criterion_transducer(
@@ -637,7 +626,6 @@

        batch_size = speech.shape[0]
        text = text[:, : text_lengths.max()]
        #print(speech.shape)
        # 1. Encoder
        encoder_out, encoder_out_chunk, encoder_out_lens = self.encode(speech, speech_lengths)

@@ -854,11 +842,6 @@
        """
        if self.criterion_transducer is None:
            try:
                # from warprnnt_pytorch import RNNTLoss
            # self.criterion_transducer = RNNTLoss(
                    # reduction="mean",
                    # fastemit_lambda=self.fastemit_lambda,
                # )
                from warp_rnnt import rnnt_loss as RNNTLoss
                self.criterion_transducer = RNNTLoss

@@ -869,12 +852,6 @@
                )
                exit(1)

        # loss_transducer = self.criterion_transducer(
        #     joint_out,
        #     target,
        #     t_len,
        #     u_len,
        # )
        log_probs = torch.log_softmax(joint_out, dim=-1)

        loss_transducer = self.criterion_transducer(

 funasr/models/predictor/cif.py

@@ -1,10 +1,12 @@
import torch

from torch import nn

from torch import Tensor

import logging

import numpy as np

from funasr.torch_utils.device_funcs import to_device

from funasr.modules.nets_utils import make_pad_mask

from funasr.modules.streaming_utils.utils import sequence_mask

from typing import Optional, Tuple



class CifPredictor(nn.Module):

    def __init__(self, idim, l_order, r_order, threshold=1.0, dropout=0.1, smooth_factor=1.0, noise_threshold=0, tail_threshold=0.45):

@@ -747,3 +749,128 @@
        predictor_alignments = index_div_bool_zeros_count_tile_out

        predictor_alignments_length = predictor_alignments.sum(-1).type(encoder_sequence_length.dtype)

        return predictor_alignments.detach(), predictor_alignments_length.detach()



class BATPredictor(nn.Module):

    def __init__(self, idim, l_order, r_order, threshold=1.0, dropout=0.1, smooth_factor=1.0, noise_threshold=0, return_accum=False):

        super(BATPredictor, self).__init__()



        self.pad = nn.ConstantPad1d((l_order, r_order), 0)

        self.cif_conv1d = nn.Conv1d(idim, idim, l_order + r_order + 1, groups=idim)

        self.cif_output = nn.Linear(idim, 1)

        self.dropout = torch.nn.Dropout(p=dropout)

        self.threshold = threshold

        self.smooth_factor = smooth_factor

        self.noise_threshold = noise_threshold

        self.return_accum = return_accum



    def cif(

        self,

        input: Tensor,

        alpha: Tensor,

        beta: float = 1.0,

        return_accum: bool = False,

    ) -> Tuple[Tensor, Tensor, Tensor, Tensor]:

        B, S, C = input.size()

        assert tuple(alpha.size()) == (B, S), f"{alpha.size()} != {(B, S)}"



        dtype = alpha.dtype

        alpha = alpha.float()



        alpha_sum = alpha.sum(1)

        feat_lengths = (alpha_sum / beta).floor().long()

        T = feat_lengths.max()



        # aggregate and integrate

        csum = alpha.cumsum(-1)

        with torch.no_grad():

            # indices used for scattering

            right_idx = (csum / beta).floor().long().clip(max=T)

            left_idx = right_idx.roll(1, dims=1)

            left_idx[:, 0] = 0



            # count # of fires from each source

            fire_num = right_idx - left_idx

            extra_weights = (fire_num - 1).clip(min=0)

            # The extra entry in last dim is for

            output = input.new_zeros((B, T + 1, C))

            source_range = torch.arange(1, 1 + S).unsqueeze(0).type_as(input)

            zero = alpha.new_zeros((1,))



        # right scatter

        fire_mask = fire_num > 0

        right_weight = torch.where(

            fire_mask,

            csum - right_idx.type_as(alpha) * beta,

            zero

        ).type_as(input)

        # assert right_weight.ge(0).all(), f"{right_weight} should be non-negative."

        output.scatter_add_(

            1,

            right_idx.unsqueeze(-1).expand(-1, -1, C),

            right_weight.unsqueeze(-1) * input

        )



        # left scatter

        left_weight = (

            alpha - right_weight - extra_weights.type_as(alpha) * beta

        ).type_as(input)

        output.scatter_add_(

            1,

            left_idx.unsqueeze(-1).expand(-1, -1, C),

            left_weight.unsqueeze(-1) * input

        )



         # extra scatters

        if extra_weights.ge(0).any():

            extra_steps = extra_weights.max().item()

            tgt_idx = left_idx

            src_feats = input * beta

            for _ in range(extra_steps):

                tgt_idx = (tgt_idx + 1).clip(max=T)

                # (B, S, 1)

                src_mask = (extra_weights > 0)

                output.scatter_add_(

                    1,

                    tgt_idx.unsqueeze(-1).expand(-1, -1, C),

                    src_feats * src_mask.unsqueeze(2)

                )

                extra_weights -= 1



        output = output[:, :T, :]



        if return_accum:

            return output, csum

        else:

            return output, alpha



    def forward(self, hidden, target_label=None, mask=None, ignore_id=-1, mask_chunk_predictor=None, target_label_length=None):

        h = hidden

        context = h.transpose(1, 2)

        queries = self.pad(context)

        memory = self.cif_conv1d(queries)

        output = memory + context

        output = self.dropout(output)

        output = output.transpose(1, 2)

        output = torch.relu(output)

        output = self.cif_output(output)

        alphas = torch.sigmoid(output)

        alphas = torch.nn.functional.relu(alphas*self.smooth_factor - self.noise_threshold)

        if mask is not None:

            alphas = alphas * mask.transpose(-1, -2).float()

        if mask_chunk_predictor is not None:

            alphas = alphas * mask_chunk_predictor

        alphas = alphas.squeeze(-1)

        if target_label_length is not None:

            target_length = target_label_length

        elif target_label is not None:

            target_length = (target_label != ignore_id).float().sum(-1)

            # logging.info("target_length: {}".format(target_length))

        else:

            target_length = None

        token_num = alphas.sum(-1)

        if target_length is not None:

            # length_noise = torch.rand(alphas.size(0), device=alphas.device) - 0.5

            # target_length = length_noise + target_length

            alphas *= ((target_length + 1e-4) / token_num)[:, None].repeat(1, alphas.size(1))

        acoustic_embeds, cif_peak = self.cif(hidden, alphas, self.threshold, self.return_accum)

        return acoustic_embeds, token_num, alphas, cif_peak


 funasr/modules/embedding.py

@@ -393,8 +393,9 @@
    def encode(self, positions: torch.Tensor = None, depth: int = None, dtype: torch.dtype = torch.float32):
        batch_size = positions.size(0)
        positions = positions.type(dtype)
        log_timescale_increment = torch.log(torch.tensor([10000], dtype=dtype)) / (depth / 2 - 1)
        inv_timescales = torch.exp(torch.arange(depth / 2).type(dtype) * (-log_timescale_increment))
        device = positions.device
        log_timescale_increment = torch.log(torch.tensor([10000], dtype=dtype, device=device)) / (depth / 2 - 1)
        inv_timescales = torch.exp(torch.arange(depth / 2, device=device).type(dtype) * (-log_timescale_increment))
        inv_timescales = torch.reshape(inv_timescales, [batch_size, -1])
        scaled_time = torch.reshape(positions, [1, -1, 1]) * torch.reshape(inv_timescales, [1, 1, -1])
        encoding = torch.cat([torch.sin(scaled_time), torch.cos(scaled_time)], dim=2)
@@ -402,7 +403,7 @@

    def forward(self, x):
        batch_size, timesteps, input_dim = x.size()
        positions = torch.arange(1, timesteps+1)[None, :]
        positions = torch.arange(1, timesteps+1, device=x.device)[None, :]
        position_encoding = self.encode(positions, input_dim, x.dtype).to(x.device)

        return x + position_encoding

 funasr/runtime/deploy_tools/funasr-runtime-deploy-offline-cpu-zh.sh

New file
@@ -0,0 +1,1450 @@
#!/usr/bin/env bash

scriptVersion="0.0.4"
scriptDate="20230702"


# Set color
RED="\033[31;1m"
GREEN="\033[32;1m"
YELLOW="\033[33;1m"
BLUE="\033[34;1m"
CYAN="\033[36;1m"
PLAIN="\033[0m"

# Info messages
DONE="${GREEN}[DONE]${PLAIN}"
ERROR="${RED}[ERROR]${PLAIN}"
WARNING="${YELLOW}[WARNING]${PLAIN}"

# Font Format
BOLD="\033[1m"
UNDERLINE="\033[4m"

# Current folder
cur_dir=`pwd`


DEFAULT_DOCKER_OFFLINE_CPU_ZH_LISTS="https://raw.githubusercontent.com/alibaba-damo-academy/FunASR/main/funasr/runtime/docs/docker_offline_cpu_zh_lists"
DEFAULT_DOCKER_IMAGE_LISTS=$DEFAULT_DOCKER_OFFLINE_CPU_ZH_LISTS
DEFAULT_FUNASR_DOCKER_URL="registry.cn-hangzhou.aliyuncs.com/funasr_repo/funasr"
DEFAULT_FUNASR_RUNTIME_RESOURCES="funasr-runtime-resources"
DEFAULT_FUNASR_LOCAL_WORKSPACE="${cur_dir}/${DEFAULT_FUNASR_RUNTIME_RESOURCES}"
DEFAULT_FUNASR_CONFIG_DIR="/var/funasr"
DEFAULT_FUNASR_CONFIG_FILE="${DEFAULT_FUNASR_CONFIG_DIR}/config"
DEFAULT_FUNASR_PROGRESS_TXT="${DEFAULT_FUNASR_CONFIG_DIR}/progress.txt"
DEFAULT_FUNASR_SERVER_LOG="${DEFAULT_FUNASR_CONFIG_DIR}/server_console.log"
DEFAULT_FUNASR_WORKSPACE_DIR="/workspace/models"
DEFAULT_DOCKER_PORT="10095"
DEFAULT_PROGRESS_FILENAME="progress.txt"
DEFAULT_SERVER_EXEC_NAME="funasr-wss-server"
DEFAULT_DOCKER_EXEC_DIR="/workspace/FunASR/funasr/runtime/websocket/build/bin"
DEFAULT_DOCKER_EXEC_PATH=${DEFAULT_DOCKER_EXEC_DIR}/${DEFAULT_SERVER_EXEC_NAME}
DEFAULT_SAMPLES_NAME="funasr_samples"
DEFAULT_SAMPLES_DIR="samples"
DEFAULT_SAMPLES_URL="https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/sample/${DEFAULT_SAMPLES_NAME}.tar.gz"

SAMPLE_CLIENTS=( \
"Python" \
"Linux_Cpp" \
)
DOCKER_IMAGES=()

# Handles the download progress bar
asr_percent_int=0
vad_percent_int=0
punc_percent_int=0
asr_title="Downloading"
asr_percent="0"
asr_speed="0KB/s"
asr_revision=""
vad_title="Downloading"
vad_percent="0"
vad_speed="0KB/s"
vad_revision=""
punc_title="Downloading"
punc_percent="0"
punc_speed="0KB/s"
punc_revision=""
serverProgress(){
    status_flag="STATUS:"
    stage=0
    wait=0
    server_status=""

    while true
    do
        if [ -f "$DEFAULT_FUNASR_PROGRESS_TXT" ]; then
            break
        else
            sleep 1
            let wait=wait+1
            if [ ${wait} -ge 6 ]; then
                break
            fi
        fi
    done

    if [ ! -f "$DEFAULT_FUNASR_PROGRESS_TXT" ]; then
        echo -e "    ${RED}The note of progress does not exist.($DEFAULT_FUNASR_PROGRESS_TXT) ${PLAIN}"
        return 98
    fi

    stage=1
    while read line
    do
        if [ $stage -eq 1 ]; then
            result=$(echo $line | grep "STATUS:")
            if [ "$result" != "" ]; then
                stage=2
                server_status=${line#*:}
                status=`expr $server_status + 0`
                if [ $status -eq 99 ]; then
                    stage=99
                fi
                continue
            fi
        elif [ $stage -eq 2 ]; then
            result=$(echo $line | grep "ASR")
            if [ "$result" != "" ]; then
                stage=3
                continue
            fi
        elif [ $stage -eq 3 ]; then
            result=$(echo $line | grep "VAD")
            if [ "$result" != "" ]; then
                stage=4
                continue
            fi
            result=$(echo $line | grep "title:")
            if [ "$result" != "" ]; then
                asr_title=${line#*:}
                continue
            fi
            result=$(echo $line | grep "percent:")
            if [ "$result" != "" ]; then
                asr_percent=${line#*:}
                continue
            fi
            result=$(echo $line | grep "speed:")
            if [ "$result" != "" ]; then
                asr_speed=${line#*:}
                continue
            fi
            result=$(echo $line | grep "revision:")
            if [ "$result" != "" ]; then
                asr_revision=${line#*:}
                continue
            fi
        elif [ $stage -eq 4 ]; then
            result=$(echo $line | grep "PUNC")
            if [ "$result" != "" ]; then
                stage=5
                continue
            fi
            result=$(echo $line | grep "title:")
            if [ "$result" != "" ]; then
                vad_title=${line#*:}
                continue
            fi
            result=$(echo $line | grep "percent:")
            if [ "$result" != "" ]; then
                vad_percent=${line#*:}
                continue
            fi
            result=$(echo $line | grep "speed:")
            if [ "$result" != "" ]; then
                vad_speed=${line#*:}
                continue
            fi
            result=$(echo $line | grep "revision:")
            if [ "$result" != "" ]; then
                vad_revision=${line#*:}
                continue
            fi
        elif [ $stage -eq 5 ]; then
            result=$(echo $line | grep "DONE")
            if [ "$result" != "" ]; then
                # Done and break.
                stage=6
                break
            fi
            result=$(echo $line | grep "title:")
            if [ "$result" != "" ]; then
                punc_title=${line#*:}
                continue
            fi
            result=$(echo $line | grep "percent:")
            if [ "$result" != "" ]; then
                punc_percent=${line#*:}
                continue
            fi
            result=$(echo $line | grep "speed:")
            if [ "$result" != "" ]; then
                punc_speed=${line#*:}
                continue
            fi
            result=$(echo $line | grep "revision:")
            if [ "$result" != "" ]; then
                punc_revision=${line#*:}
                continue
            fi
        elif [ $stage -eq 99 ]; then
            echo -e "    ${RED}ERROR: $line${PLAIN}"
        fi
    done < $DEFAULT_FUNASR_PROGRESS_TXT

    if [ $stage -ne 99 ]; then
        drawProgress "ASR " $asr_title $asr_percent $asr_speed $asr_revision $asr_percent_int
        asr_percent_int=$?
        drawProgress "VAD " $vad_title $vad_percent $vad_speed $vad_revision $vad_percent_int
        vad_percent_int=$?
        drawProgress "PUNC" $punc_title $punc_percent $punc_speed $punc_revision $punc_percent_int
        punc_percent_int=$?
    fi

    return $stage
}

drawProgress(){
    model=$1
    title=$2
    percent_str=$3
    speed=$4
    revision=$5
    latest_percent=$6

    progress=0
    if [ ! -z "$percent_str" ]; then
        progress=`expr $percent_str + 0`
        latest_percent=`expr $latest_percent + 0`
        if [ $progress -ne 0 ] && [ $progress -lt $latest_percent ]; then
            progress=$latest_percent
        fi
    fi

    loading_flag="Loading"
    if [ "$title" = "$loading_flag" ]; then
        progress=100
    fi

    i=0
    str=""
    let max=progress/2
    while [ $i -lt $max ]
    do
        let i++
        str+='='
    done
    let color=36
    let index=max*2
    if [ -z "$speed" ]; then
        printf "\r    \e[0;${CYAN}[%s][%-11s][%-50s][%d%%][%s]\e[0m" "$model" "$title" "$str" "$$index" "$revision"
    else
        printf "\r    \e[0;${CYAN}[%s][%-11s][%-50s][%3d%%][%8s][%s]\e[0m" "$model" "$title" "$str" "$index" "$speed" "$revision"
    fi
    printf "\n"

    return $progress
}

menuSelection(){
    local menu
    menu=($(echo "$@"))
    result=1
    show_no=1
    menu_no=0
    len=${#menu[@]}

    while true
    do
        echo -e "    ${BOLD}${show_no})${PLAIN} ${menu[menu_no]}"

        let show_no++
        let menu_no++
        if [ $menu_no -ge $len ]; then
            break
        fi
    done

    while true
    do
        echo -e "  Enter your choice, default(${CYAN}1${PLAIN}): \c"
        read result
        if [ -z "$result" ]; then
            result=1
        fi

        expr $result + 0 &>/dev/null
        if [ $? -eq 0 ]; then
            if [ $result -ge 1 ] && [ $result -le $len ]; then
                break
            else
                echo -e "    ${RED}Input error, please input correct number!${PLAIN}"
            fi
        else
            echo -e "    ${RED}Input error, please input correct number!${PLAIN}"
        fi
    done
    
    return $result
}


full_path=""
relativePathToFullPath(){
    relativePath=$1
    firstChar=${relativePath: 0: 1}
    if [[ "$firstChar" == "" ]]; then
        full_path=$relativePath
    elif [[ "$firstChar" == "/" ]]; then
        full_path=$relativePath
    fi

    tmpPath1=`dirname $relativePath`
    tmpFullpath1=`cd $tmpPath1 && pwd`
    tmpPath2=`basename $relativePath`
    full_path=${tmpFullpath1}/${tmpPath2}
}

initConfiguration(){
    if [ ! -z "$DEFAULT_FUNASR_CONFIG_DIR" ]; then
        mkdir -p $DEFAULT_FUNASR_CONFIG_DIR
    fi
    if [ ! -f $DEFAULT_FUNASR_CONFIG_FILE ]; then
        touch $DEFAULT_FUNASR_CONFIG_FILE
    fi
}

initParameters(){
    # Init workspace in local by new parameters.
    PARAMS_FUNASR_SAMPLES_LOCAL_PATH=${PARAMS_FUNASR_LOCAL_WORKSPACE}/${DEFAULT_SAMPLES_NAME}.tar.gz
    PARAMS_FUNASR_SAMPLES_LOCAL_DIR=${PARAMS_FUNASR_LOCAL_WORKSPACE}/${DEFAULT_SAMPLES_DIR}
    PARAMS_FUNASR_LOCAL_MODELS_DIR="${PARAMS_FUNASR_LOCAL_WORKSPACE}/models"

    if [ ! -z "$PARAMS_FUNASR_LOCAL_WORKSPACE" ]; then
        mkdir -p $PARAMS_FUNASR_LOCAL_WORKSPACE
    fi
    if [ ! -z "$PARAMS_FUNASR_LOCAL_MODELS_DIR" ]; then
        mkdir -p $PARAMS_FUNASR_LOCAL_MODELS_DIR
    fi
}

# Parse the parameters from the docker list file.
docker_info_cur_key=""
docker_info_cur_val=""
findTypeOfDockerInfo(){
    line=$1
    result=$(echo $line | grep ":")
    if [ "$result" != "" ]; then
        docker_info_cur_key=$result
        docker_info_cur_val=""
    else
        docker_info_cur_val=$(echo $line)
    fi
}

# Get a list of docker images.
readDockerInfoFromUrl(){
    list_url=$DEFAULT_DOCKER_IMAGE_LISTS
    while true
    do
        content=$(curl --connect-timeout 10 -m 10 -s $list_url)
        if [ ! -z "$content" ]; then
            break
        else
            echo -e "    ${RED}Unable to get docker image list due to network issues, try again.${PLAIN}"
        fi
    done
    array=($(echo "$content"))
    len=${#array[@]}

    stage=0
    docker_flag="DOCKER:"
    judge_flag=":"
    for i in ${array[@]}
    do
        findTypeOfDockerInfo $i
        if [ "$docker_info_cur_key" = "DOCKER:" ]; then
            if [ ! -z "$docker_info_cur_val" ]; then
                docker_name=${DEFAULT_FUNASR_DOCKER_URL}:${docker_info_cur_val}
                DOCKER_IMAGES[${#DOCKER_IMAGES[*]}]=$docker_name 
            fi
        elif [ "$docker_info_cur_key" = "DEFAULT_ASR_MODEL:" ]; then
            if [ ! -z "$docker_info_cur_val" ]; then
                PARAMS_ASR_ID=$docker_info_cur_val
            fi
        elif [ "$docker_info_cur_key" = "DEFAULT_VAD_MODEL:" ]; then
            if [ ! -z "$docker_info_cur_val" ]; then
                PARAMS_VAD_ID=$docker_info_cur_val
            fi
        elif [ "$docker_info_cur_key" = "DEFAULT_PUNC_MODEL:" ]; then
            if [ ! -z "$docker_info_cur_val" ]; then
                PARAMS_PUNC_ID=$docker_info_cur_val
            fi
        fi
    done
    echo -e "    $DONE"
}

# Make sure root user.
rootNess(){
    echo -e "${UNDERLINE}${BOLD}[0/5]${PLAIN}"
    echo -e "  ${YELLOW}Please check root access.${PLAIN}"

    echo -e "    ${WARNING} MUST RUN AS ${RED}ROOT${PLAIN} USER!"
    if [[ $EUID -ne 0 ]]; then
        echo -e "  ${ERROR} MUST RUN AS ${RED}ROOT${PLAIN} USER!"
    fi

    cd $cur_dir
    echo
}

# Get a list of docker images and select them.
selectDockerImages(){
    echo -e "${UNDERLINE}${BOLD}[1/5]${PLAIN}"
    echo -e "  ${YELLOW}Getting the list of docker images, please wait a few seconds.${PLAIN}"
    readDockerInfoFromUrl
    echo

    echo -e "  ${YELLOW}Please choose the Docker image.${PLAIN}"
    menuSelection ${DOCKER_IMAGES[*]}
    result=$?
    index=`expr $result - 1`

    PARAMS_DOCKER_IMAGE=${DOCKER_IMAGES[${index}]}
    echo -e "  ${UNDERLINE}You have chosen the Docker image:${PLAIN} ${GREEN}${PARAMS_DOCKER_IMAGE}${PLAIN}"

    checkDockerExist
    result=$?
    result=`expr $result + 0`
    if [ ${result} -eq 50 ]; then
        return 50
    fi

    echo
}

# Configure FunASR server host port setting.
setupHostPort(){
    echo -e "${UNDERLINE}${BOLD}[2/5]${PLAIN}"

    params_host_port=`sed '/^PARAMS_HOST_PORT=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ -z "$params_host_port" ]; then
        PARAMS_HOST_PORT="10095"
    else
        PARAMS_HOST_PORT=$params_host_port
    fi

    while true
    do
        echo -e "  ${YELLOW}Please input the opened port in the host used for FunASR server.${PLAIN}"
        echo -e "  Setting the opened host port [1-65535], default(${CYAN}${PARAMS_HOST_PORT}${PLAIN}): \c"
        read PARAMS_HOST_PORT

        if [ -z "$PARAMS_HOST_PORT" ]; then
            params_host_port=`sed '/^PARAMS_HOST_PORT=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
            if [ -z "$params_host_port" ]; then
                PARAMS_HOST_PORT="10095"
            else
                PARAMS_HOST_PORT=$params_host_port
            fi
        fi
        expr $PARAMS_HOST_PORT + 0 &>/dev/null
        if [ $? -eq 0 ]; then
            if [ $PARAMS_HOST_PORT -ge 1 ] && [ $PARAMS_HOST_PORT -le 65535 ]; then
                echo -e "  ${UNDERLINE}The port of the host is${PLAIN} ${GREEN}${PARAMS_HOST_PORT}${PLAIN}"
                echo -e "  ${UNDERLINE}The port in Docker for FunASR server is${PLAIN} ${GREEN}${PARAMS_DOCKER_PORT}${PLAIN}"
                break
            else
                echo -e "  ${RED}Input error, please input correct number!${PLAIN}"
            fi
        else
            echo -e "  ${RED}Input error, please input correct number!${PLAIN}"
        fi
    done
    echo
}

complementParameters(){
    # parameters about ASR model
    if [ ! -z "$PARAMS_ASR_ID" ]; then
        PARAMS_DOCKER_ASR_PATH=${PARAMS_DOWNLOAD_MODEL_DIR}/${PARAMS_ASR_ID}
        PARAMS_DOCKER_ASR_DIR=$(dirname "$PARAMS_DOCKER_ASR_PATH")
        PARAMS_LOCAL_ASR_PATH=${PARAMS_FUNASR_LOCAL_MODELS_DIR}/${PARAMS_ASR_ID}
        PARAMS_LOCAL_ASR_DIR=$(dirname "$PARAMS_LOCAL_ASR_PATH")
    fi

    # parameters about VAD model
    if [ ! -z "$PARAMS_VAD_ID" ]; then
            PARAMS_DOCKER_VAD_PATH=${PARAMS_DOWNLOAD_MODEL_DIR}/${PARAMS_VAD_ID}
            PARAMS_DOCKER_VAD_DIR=$(dirname "$PARAMS_DOCKER_VAD_PATH")
            PARAMS_LOCAL_VAD_PATH=${PARAMS_FUNASR_LOCAL_MODELS_DIR}/${PARAMS_VAD_ID}
            PARAMS_LOCAL_VAD_DIR=$(dirname "$PARAMS_LOCAL_VAD_PATH")
    fi

    # parameters about PUNC model
    if [ ! -z "$PARAMS_PUNC_ID" ]; then
        PARAMS_DOCKER_PUNC_PATH=${PARAMS_DOWNLOAD_MODEL_DIR}/${PARAMS_PUNC_ID}
        PARAMS_DOCKER_PUNC_DIR=$(dirname "${PARAMS_DOCKER_PUNC_PATH}")
        PARAMS_LOCAL_PUNC_PATH=${PARAMS_FUNASR_LOCAL_MODELS_DIR}/${PARAMS_PUNC_ID}
        PARAMS_LOCAL_PUNC_DIR=$(dirname "${PARAMS_LOCAL_PUNC_PATH}")
    fi

    # parameters about thread_num
    params_decoder_thread_num=`sed '/^PARAMS_DECODER_THREAD_NUM=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ -z "$params_decoder_thread_num" ]; then
        PARAMS_DECODER_THREAD_NUM=$CPUNUM
    else
        PARAMS_DECODER_THREAD_NUM=$params_decoder_thread_num
    fi

    multiple_io=4
    PARAMS_DECODER_THREAD_NUM=`expr ${PARAMS_DECODER_THREAD_NUM} + 0`
    PARAMS_IO_THREAD_NUM=`expr ${PARAMS_DECODER_THREAD_NUM} / ${multiple_io}`
    if [ $PARAMS_IO_THREAD_NUM -eq 0 ]; then
        PARAMS_IO_THREAD_NUM=1
    fi
}

paramsFromDefault(){
    initConfiguration

    echo -e "  ${YELLOW}Load parameters from${PLAIN} ${GREEN}${DEFAULT_FUNASR_CONFIG_FILE}${PLAIN}"
    echo

    funasr_local_workspace=`sed '/^PARAMS_FUNASR_LOCAL_WORKSPACE=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$funasr_local_workspace" ]; then
        PARAMS_FUNASR_LOCAL_WORKSPACE=$funasr_local_workspace
    fi
    funasr_samples_local_dir=`sed '/^PARAMS_FUNASR_SAMPLES_LOCAL_DIR=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$funasr_samples_local_dir" ]; then
        PARAMS_FUNASR_SAMPLES_LOCAL_DIR=$funasr_samples_local_dir
    fi
    funasr_samples_local_path=`sed '/^PARAMS_FUNASR_SAMPLES_LOCAL_PATH=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$funasr_samples_local_path" ]; then
        PARAMS_FUNASR_SAMPLES_LOCAL_PATH=$funasr_samples_local_path
    fi
    funasr_local_models_dir=`sed '/^PARAMS_FUNASR_LOCAL_MODELS_DIR=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$funasr_local_models_dir" ]; then
        PARAMS_FUNASR_LOCAL_MODELS_DIR=$funasr_local_models_dir
    fi
    funasr_config_path=`sed '/^PARAMS_FUNASR_CONFIG_PATH=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$funasr_config_path" ]; then
        PARAMS_FUNASR_CONFIG_PATH=$funasr_config_path
    fi

    docker_image=`sed '/^PARAMS_DOCKER_IMAGE=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$docker_image" ]; then
        PARAMS_DOCKER_IMAGE=$docker_image
    fi
    download_model_dir=`sed '/^PARAMS_DOWNLOAD_MODEL_DIR=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$download_model_dir" ]; then
        PARAMS_DOWNLOAD_MODEL_DIR=$download_model_dir
    fi
    PARAMS_LOCAL_ASR_PATH=`sed '/^PARAMS_LOCAL_ASR_PATH=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$local_asr_path" ]; then
        PARAMS_LOCAL_ASR_PATH=$local_asr_path
    fi
    docker_asr_path=`sed '/^PARAMS_DOCKER_ASR_PATH=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$docker_asr_path" ]; then
        PARAMS_DOCKER_ASR_PATH=$docker_asr_path
    fi
    asr_id=`sed '/^PARAMS_ASR_ID=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$asr_id" ]; then
        PARAMS_ASR_ID=$asr_id
    fi
    local_vad_path=`sed '/^PARAMS_LOCAL_VAD_PATH=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$local_vad_path" ]; then
        PARAMS_LOCAL_VAD_PATH=$local_vad_path
    fi
    docker_vad_path=`sed '/^PARAMS_DOCKER_VAD_PATH=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$docker_vad_path" ]; then
        PARAMS_DOCKER_VAD_PATH=$docker_vad_path
    fi
    vad_id=`sed '/^PARAMS_VAD_ID=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$vad_id" ]; then
        PARAMS_VAD_ID=$vad_id
    fi
    local_punc_path=`sed '/^PARAMS_LOCAL_PUNC_PATH=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$local_punc_path" ]; then
        PARAMS_LOCAL_PUNC_PATH=$local_punc_path
    fi
    docker_punc_path=`sed '/^PARAMS_DOCKER_PUNC_PATH=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$docker_punc_path" ]; then
        PARAMS_DOCKER_PUNC_PATH=$docker_punc_path
    fi
    punc_id=`sed '/^PARAMS_PUNC_ID=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$punc_id" ]; then
        PARAMS_PUNC_ID=$punc_id
    fi
    docker_exec_path=`sed '/^PARAMS_DOCKER_EXEC_PATH=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$docker_exec_path" ]; then
        PARAMS_DOCKER_EXEC_PATH=$docker_exec_path
    fi
    host_port=`sed '/^PARAMS_HOST_PORT=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$host_port" ]; then
        PARAMS_HOST_PORT=$host_port
    fi
    docker_port=`sed '/^PARAMS_DOCKER_PORT=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$docker_port" ]; then
        PARAMS_DOCKER_PORT=$docker_port
    fi
    decode_thread_num=`sed '/^PARAMS_DECODER_THREAD_NUM=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$decode_thread_num" ]; then
        PARAMS_DECODER_THREAD_NUM=$decode_thread_num
    fi
    io_thread_num=`sed '/^PARAMS_IO_THREAD_NUM=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
    if [ ! -z "$io_thread_num" ]; then
        PARAMS_IO_THREAD_NUM=$io_thread_num
    fi
}

saveParams(){
    echo "$i" > $DEFAULT_FUNASR_CONFIG_FILE
    echo -e "  ${GREEN}Parameters are stored in the file ${DEFAULT_FUNASR_CONFIG_FILE}${PLAIN}"

    echo "PARAMS_DOCKER_IMAGE=${PARAMS_DOCKER_IMAGE}" > $DEFAULT_FUNASR_CONFIG_FILE

    echo "PARAMS_FUNASR_LOCAL_WORKSPACE=${PARAMS_FUNASR_LOCAL_WORKSPACE}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_FUNASR_SAMPLES_LOCAL_DIR=${PARAMS_FUNASR_SAMPLES_LOCAL_DIR}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_FUNASR_SAMPLES_LOCAL_PATH=${PARAMS_FUNASR_SAMPLES_LOCAL_PATH}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_FUNASR_LOCAL_MODELS_DIR=${PARAMS_FUNASR_LOCAL_MODELS_DIR}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_FUNASR_CONFIG_PATH=${PARAMS_FUNASR_CONFIG_PATH}" >> $DEFAULT_FUNASR_CONFIG_FILE

    echo "PARAMS_DOWNLOAD_MODEL_DIR=${PARAMS_DOWNLOAD_MODEL_DIR}" >> $DEFAULT_FUNASR_CONFIG_FILE

    echo "PARAMS_DOCKER_EXEC_PATH=${PARAMS_DOCKER_EXEC_PATH}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_DOCKER_EXEC_DIR=${PARAMS_DOCKER_EXEC_DIR}" >> $DEFAULT_FUNASR_CONFIG_FILE

    echo "PARAMS_LOCAL_ASR_PATH=${PARAMS_LOCAL_ASR_PATH}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_LOCAL_ASR_DIR=${PARAMS_LOCAL_ASR_DIR}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_DOCKER_ASR_PATH=${PARAMS_DOCKER_ASR_PATH}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_DOCKER_ASR_DIR=${PARAMS_DOCKER_ASR_DIR}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_ASR_ID=${PARAMS_ASR_ID}" >> $DEFAULT_FUNASR_CONFIG_FILE

    echo "PARAMS_LOCAL_PUNC_PATH=${PARAMS_LOCAL_PUNC_PATH}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_LOCAL_PUNC_DIR=${PARAMS_LOCAL_PUNC_DIR}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_DOCKER_PUNC_PATH=${PARAMS_DOCKER_PUNC_PATH}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_DOCKER_PUNC_DIR=${PARAMS_DOCKER_PUNC_DIR}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_PUNC_ID=${PARAMS_PUNC_ID}" >> $DEFAULT_FUNASR_CONFIG_FILE

    echo "PARAMS_LOCAL_VAD_PATH=${PARAMS_LOCAL_VAD_PATH}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_LOCAL_VAD_DIR=${PARAMS_LOCAL_VAD_DIR}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_DOCKER_VAD_PATH=${PARAMS_DOCKER_VAD_PATH}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_DOCKER_VAD_DIR=${PARAMS_DOCKER_VAD_DIR}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_VAD_ID=${PARAMS_VAD_ID}" >> $DEFAULT_FUNASR_CONFIG_FILE

    echo "PARAMS_HOST_PORT=${PARAMS_HOST_PORT}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_DOCKER_PORT=${PARAMS_DOCKER_PORT}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_DECODER_THREAD_NUM=${PARAMS_DECODER_THREAD_NUM}" >> $DEFAULT_FUNASR_CONFIG_FILE
    echo "PARAMS_IO_THREAD_NUM=${PARAMS_IO_THREAD_NUM}" >> $DEFAULT_FUNASR_CONFIG_FILE
}

showAllParams(){
    echo -e "${UNDERLINE}${BOLD}[3/5]${PLAIN}"
    echo -e "  ${YELLOW}Show parameters of FunASR server setting and confirm to run ...${PLAIN}"
    echo

    if [ ! -z "$PARAMS_DOCKER_IMAGE" ]; then
        echo -e "  The current Docker image is                                    : ${GREEN}${PARAMS_DOCKER_IMAGE}${PLAIN}"
    fi
    if [ ! -z "$PARAMS_FUNASR_LOCAL_WORKSPACE" ]; then
        echo -e "  The local workspace path is                                    : ${GREEN}${PARAMS_FUNASR_LOCAL_WORKSPACE}${PLAIN}"
    fi

    if [ ! -z "$PARAMS_DOWNLOAD_MODEL_DIR" ]; then
        echo -e "  The model will be automatically downloaded in Docker           : ${GREEN}${PARAMS_DOWNLOAD_MODEL_DIR}${PLAIN}"
    fi
    echo

    if [ ! -z "$PARAMS_ASR_ID" ]; then
        echo -e "  The ASR model_id used                                          : ${GREEN}${PARAMS_ASR_ID}${PLAIN}"
    fi
    if [ ! -z "$PARAMS_LOCAL_ASR_PATH" ]; then
        echo -e "  The path to the local ASR model directory for the load         : ${GREEN}${PARAMS_LOCAL_ASR_PATH}${PLAIN}"
    fi
    echo -e "  The ASR model directory corresponds to the directory in Docker : ${GREEN}${PARAMS_DOCKER_ASR_PATH}${PLAIN}"

    if [ ! -z "$PARAMS_VAD_ID" ]; then
        echo -e "  The VAD model_id used                                          : ${GREEN}${PARAMS_VAD_ID}${PLAIN}"
    fi
    if [ ! -z "$PARAMS_LOCAL_VAD_PATH" ]; then
        echo -e "  The path to the local VAD model directory for the load         : ${GREEN}${PARAMS_LOCAL_VAD_PATH}${PLAIN}"
    fi
    echo -e "  The VAD model directory corresponds to the directory in Docker : ${GREEN}${PARAMS_DOCKER_VAD_PATH}${PLAIN}"

    if [ ! -z "$PARAMS_PUNC_ID" ]; then
        echo -e "  The PUNC model_id used                                         : ${GREEN}${PARAMS_PUNC_ID}${PLAIN}"
    fi
    if [ ! -z "$PARAMS_LOCAL_PUNC_PATH" ]; then
        echo -e "  The path to the local PUNC model directory for the load        : ${GREEN}${PARAMS_LOCAL_PUNC_PATH}${PLAIN}"
    fi
    echo -e "  The PUNC model directory corresponds to the directory in Docker: ${GREEN}${PARAMS_DOCKER_PUNC_PATH}${PLAIN}"
    echo

    echo -e "  The path in the docker of the FunASR service executor          : ${GREEN}${PARAMS_DOCKER_EXEC_PATH}${PLAIN}"

    echo -e "  Set the host port used for use by the FunASR service           : ${GREEN}${PARAMS_HOST_PORT}${PLAIN}"
    echo -e "  Set the docker port used by the FunASR service                 : ${GREEN}${PARAMS_DOCKER_PORT}${PLAIN}"

    echo -e "  Set the number of threads used for decoding the FunASR service : ${GREEN}${PARAMS_DECODER_THREAD_NUM}${PLAIN}"
    echo -e "  Set the number of threads used for IO the FunASR service       : ${GREEN}${PARAMS_IO_THREAD_NUM}${PLAIN}"
    echo

    if [ ! -z "$PARAMS_FUNASR_SAMPLES_LOCAL_DIR" ]; then
        echo -e "  Sample code will be store in local                             : ${GREEN}${PARAMS_FUNASR_SAMPLES_LOCAL_DIR}${PLAIN}"
    fi

    echo
    while true
    do
        params_confirm="y"
        echo -e "  ${YELLOW}Please input [Y/n] to confirm the parameters.${PLAIN}"
        echo -e "  [y] Verify that these parameters are correct and that the service will run."
        echo -e "  [n] The parameters set are incorrect, it will be rolled out, please rerun."
        echo -e "  read confirmation[${CYAN}Y${PLAIN}/n]: \c"
        read params_confirm

        if [ -z "$params_confirm" ]; then
            params_confirm="y"
        fi
        YES="Y"
        yes="y"
        NO="N"
        no="n"
        echo
        if [ "$params_confirm" = "$YES" ] || [ "$params_confirm" = "$yes" ]; then
            echo -e "  ${GREEN}Will run FunASR server later ...${PLAIN}"
            break
        elif [ "$params_confirm" = "$NO" ] || [ "$params_confirm" = "$no" ]; then
            echo -e "  ${RED}The parameters set are incorrect, please rerun ...${PLAIN}"
            exit 1
        else
            echo "again ..."
        fi
    done

    saveParams
    echo
    sleep 1
}

# Install docker
installFunasrDocker(){
    echo -e "${UNDERLINE}${BOLD}[4/5]${PLAIN}"

    if [ $DOCKERINFOLEN -gt 30 ]; then
        echo -e "  ${YELLOW}Docker has installed.${PLAIN}"
    else
        lowercase_osid=$(echo ${OSID} | tr '[A-Z]' '[a-z]')
        echo -e "  ${YELLOW}Start install docker for ${lowercase_osid} ${PLAIN}"
        DOCKER_INSTALL_CMD="curl -fsSL https://get.docker.com | bash -s docker --mirror Aliyun"
        DOCKER_INSTALL_RUN_CMD=""

        case "$lowercase_osid" in
            ubuntu)
                DOCKER_INSTALL_CMD="curl -fsSL https://test.docker.com -o test-docker.sh"
                DOCKER_INSTALL_RUN_CMD="sudo sh test-docker.sh"
                ;;
            centos)
                DOCKER_INSTALL_CMD="curl -fsSL https://get.docker.com | bash -s docker --mirror Aliyun"
                ;;
            debian)
                DOCKER_INSTALL_CMD="curl -fsSL https://get.docker.com -o get-docker.sh"
                DOCKER_INSTALL_RUN_CMD="sudo sh get-docker.sh"
                ;;
            \"alios\")
                DOCKER_INSTALL_CMD="curl -fsSL https://get.docker.com -o get-docker.sh"
                DOCKER_INSTALL_RUN_CMD="sudo sh get-docker.sh"
                ;;
            \"alinux\")
                DOCKER_INSTALL_CMD="sudo yum -y install dnf"
                DOCKER_INSTALL_RUN_CMD="sudo dnf -y install docker"
                ;;
            *)
                echo -e "  ${RED}$lowercase_osid is not supported.${PLAIN}"
                ;;
        esac

        echo -e "  Get docker installer: ${GREEN}${DOCKER_INSTALL_CMD}${PLAIN}"
        echo -e "  Get docker run: ${GREEN}${DOCKER_INSTALL_RUN_CMD}${PLAIN}"

        $DOCKER_INSTALL_CMD
        if [ ! -z "$DOCKER_INSTALL_RUN_CMD" ]; then
            $DOCKER_INSTALL_RUN_CMD
        fi
        sudo systemctl start docker

        DOCKERINFO=$(sudo docker info | wc -l)
        DOCKERINFOLEN=`expr ${DOCKERINFO} + 0`
        if [ $DOCKERINFOLEN -gt 30 ]; then
            echo -e "  ${GREEN}Docker install success, start docker server.${PLAIN}"
            sudo systemctl start docker
        else
            echo -e "  ${RED}Docker install failed!${PLAIN}"
            exit 1
        fi
    fi

    echo
    sleep 1

    # Download docker image
    echo -e "  ${YELLOW}Pull docker image(${PARAMS_DOCKER_IMAGE})...${PLAIN}"

    sudo docker pull $PARAMS_DOCKER_IMAGE

    echo
    sleep 1
}

dockerRun(){
    echo -e "${UNDERLINE}${BOLD}[5/5]${PLAIN}"
    echo -e "  ${YELLOW}Construct command and run docker ...${PLAIN}"

    run_cmd="sudo docker run"
    port_map=" -p ${PARAMS_HOST_PORT}:${PARAMS_DOCKER_PORT}"
    dir_params=" --privileged=true"
    dir_map_params=""
    if [ ! -z "$PARAMS_LOCAL_ASR_DIR" ]; then
        if [ -z "$dir_map_params" ]; then
            dir_map_params="${dir_params} -v ${PARAMS_LOCAL_ASR_DIR}:${PARAMS_DOCKER_ASR_DIR}"
        else
            dir_map_params="${dir_map_params} -v ${PARAMS_LOCAL_ASR_DIR}:${PARAMS_DOCKER_ASR_DIR}"
        fi
    fi
    if [ ! -z "$PARAMS_LOCAL_VAD_DIR" ]; then
        if [ -z "$dir_map_params" ]; then
            dir_map_params="${dir_params} -v ${PARAMS_LOCAL_VAD_DIR}:${PARAMS_DOCKER_VAD_DIR}"
        else
            dir_map_params="${dir_map_params} -v ${PARAMS_LOCAL_VAD_DIR}:${PARAMS_DOCKER_VAD_DIR}"
        fi
    fi
    if [ ! -z "$PARAMS_LOCAL_PUNC_DIR" ]; then
        if [ -z "$dir_map_params" ]; then
            dir_map_params="${dir_params} -v ${PARAMS_LOCAL_PUNC_DIR}:${PARAMS_DOCKER_PUNC_DIR}"
        else
            dir_map_params="${dir_map_params} -v ${PARAMS_LOCAL_VAD_DIR}:${PARAMS_DOCKER_VAD_DIR}"
        fi
    fi

    exec_params="\"exec\":\"${PARAMS_DOCKER_EXEC_PATH}\""
    if [ ! -z "$PARAMS_ASR_ID" ]; then
        asr_params="\"--model-dir\":\"${PARAMS_ASR_ID}\""
    else
        asr_params="\"--model-dir\":\"${PARAMS_DOCKER_ASR_PATH}\""
    fi
    if [ ! -z "$PARAMS_VAD_ID" ]; then
        vad_params="\"--vad-dir\":\"${PARAMS_VAD_ID}\""
    else
        vad_params="\"--vad-dir\":\"${PARAMS_DOCKER_VAD_PATH}\""
    fi
    if [ ! -z "$PARAMS_PUNC_ID" ]; then
        punc_params="\"--punc-dir\":\"${PARAMS_PUNC_ID}\""
    else
        punc_params="\"--punc-dir\":\"${PARAMS_DOCKER_PUNC_PATH}\""
    fi
    download_params="\"--download-model-dir\":\"${PARAMS_DOWNLOAD_MODEL_DIR}\""
    if [ -z "$PARAMS_DOWNLOAD_MODEL_DIR" ]; then
        model_params="${asr_params},${vad_params},${punc_params}"
    else
        model_params="${asr_params},${vad_params},${punc_params},${download_params}"
    fi

    decoder_params="\"--decoder-thread-num\":\"${PARAMS_DECODER_THREAD_NUM}\""
    io_params="\"--io-thread-num\":\"${PARAMS_IO_THREAD_NUM}\""
    thread_params=${decoder_params},${io_params}
    port_params="\"--port\":\"${PARAMS_DOCKER_PORT}\""
    crt_path="\"--certfile\":\"/workspace/FunASR/funasr/runtime/ssl_key/server.crt\""
    key_path="\"--keyfile\":\"/workspace/FunASR/funasr/runtime/ssl_key/server.key\""

    env_params=" -v ${DEFAULT_FUNASR_CONFIG_DIR}:/workspace/.config"
    env_params=" ${env_params} --env DAEMON_SERVER_CONFIG={\"server\":[{${exec_params},${model_params},${thread_params},${port_params},${crt_path},${key_path}}]}"

    run_cmd="${run_cmd}${port_map}${dir_map_params}${env_params}"
    run_cmd="${run_cmd} -it -d ${PARAMS_DOCKER_IMAGE}"

    # check Docker
    checkDockerExist
    result=$?
    result=`expr ${result} + 0`
    if [ ${result} -eq 50 ]; then
        return 50
    fi

    rm -f ${DEFAULT_FUNASR_PROGRESS_TXT}
    rm -f ${DEFAULT_FUNASR_SERVER_LOG}

    $run_cmd

    echo
    echo -e "  ${YELLOW}Loading models:${PLAIN}"

    # Hide the cursor, start draw progress.
    printf "\e[?25l"
    while true
    do
        serverProgress
        result=$?
        stage=`expr ${result} + 0`
        if [ ${stage} -eq 0 ]; then
            break
        elif [ ${stage} -gt 0 ] && [ ${stage} -lt 6 ]; then
            sleep 0.1
            # clear 3 lines
            printf "\033[3A"
        elif [ ${stage} -eq 6 ]; then
            break
        elif [ ${stage} -eq 98 ]; then
            return 98
        else
            echo -e "  ${RED}Starting FunASR server failed.${PLAIN}"
            echo
            # Display the cursor
            printf "\e[?25h"
            return 99
        fi
    done
    # Display the cursor
    printf "\e[?25h"

    echo -e "  ${GREEN}The service has been started.${PLAIN}"
    echo

    deploySamples
    echo -e "  ${BOLD}The sample code is already stored in the ${PLAIN}(${GREEN}${PARAMS_FUNASR_SAMPLES_LOCAL_DIR}${PLAIN}) ."
    echo -e "  ${BOLD}If you want to see an example of how to use the client, you can run ${PLAIN}${GREEN}sudo bash funasr-runtime-deploy-offline-cpu-zh.sh client${PLAIN} ."
    echo
}

installPythonDependencyForPython(){
    echo -e "${YELLOW}Install Python dependent environments ...${PLAIN}"

    echo -e "  Export dependency of Cpp sample."
    pre_cmd="export LD_LIBRARY_PATH=${PARAMS_FUNASR_SAMPLES_LOCAL_DIR}/cpp/libs:\$LD_LIBRARY_PATH"
    $pre_cmd
    echo

    echo -e "  Install requirements of Python sample."
    pre_cmd="pip3 install click>=8.0.4"
    $pre_cmd
    echo

    pre_cmd="pip3 install -r ${PARAMS_FUNASR_SAMPLES_LOCAL_DIR}/python/requirements_client.txt"
    echo -e "  Run ${BLUE}${pre_cmd}${PLAIN}"
    $pre_cmd
    echo

    lowercase_osid=$(echo ${OSID} | tr '[A-Z]' '[a-z]')
    case "$lowercase_osid" in
        ubuntu)
            pre_cmd="sudo apt-get install -y ffmpeg"
            ;;
        centos)
            pre_cmd="sudo yum install -y ffmpeg"
            ;;
        debian)
            pre_cmd="sudo apt-get install -y ffmpeg"
            ;;
        \"alios\")
            pre_cmd="sudo yum install -y ffmpeg"
            ;;
        \"alinux\")
            pre_cmd="sudo yum install -y ffmpeg"
            ;;
        *)
            echo -e "  ${RED}$lowercase_osid is not supported.${PLAIN}"
            ;;
    esac
    echo -e "  Run ${BLUE}${pre_cmd}${PLAIN}"
    echo

    pre_cmd="pip3 install ffmpeg-python"
    echo -e "  Run ${BLUE}${pre_cmd}${PLAIN}"
    $pre_cmd
    echo
}

deploySamples(){
    if [ ! -d $PARAMS_FUNASR_SAMPLES_LOCAL_DIR ]; then
        echo -e "${YELLOW}Downloading samples to${PLAIN} ${CYAN}${PARAMS_FUNASR_LOCAL_WORKSPACE}${PLAIN} ${YELLOW}...${PLAIN}"

        download_cmd="curl ${DEFAULT_SAMPLES_URL} -o ${PARAMS_FUNASR_SAMPLES_LOCAL_PATH}"
        untar_cmd="tar -zxf ${PARAMS_FUNASR_SAMPLES_LOCAL_PATH} -C ${PARAMS_FUNASR_LOCAL_WORKSPACE}"

        if [ ! -f "$PARAMS_FUNASR_SAMPLES_LOCAL_PATH" ]; then
            $download_cmd
        fi
        $untar_cmd
        echo

        installPythonDependencyForPython
        echo
    fi
}

checkDockerExist(){
    result=$(sudo docker ps | grep ${PARAMS_DOCKER_IMAGE} | wc -l)
    result=`expr ${result} + 0`
    if [ ${result} -ne 0 ]; then
        echo
        echo -e "  ${RED}Docker: ${PARAMS_DOCKER_IMAGE} has been launched, please run (${PLAIN}${GREEN}sudo bash funasr-runtime-deploy-offline-cpu-zh.sh stop${PLAIN}${RED}) to stop Docker first.${PLAIN}"
        return 50
    fi
}

dockerExit(){
    echo -e "  ${YELLOW}Stop docker(${PLAIN}${GREEN}${PARAMS_DOCKER_IMAGE}${PLAIN}${YELLOW}) server ...${PLAIN}"
    sudo docker stop `sudo docker ps -a| grep ${PARAMS_DOCKER_IMAGE} | awk '{print $1}' `
    echo
    sleep 1
}

modelChange(){
    model_type=$1
    model_id=$2
    local_flag=0

    if [ -d "$model_id" ]; then
        local_flag=1
    else
        local_flag=0
    fi

    result=$(echo $model_type | grep "--asr_model")
    if [ "$result" != "" ]; then
        if [ $local_flag -eq 0 ]; then
            PARAMS_ASR_ID=$model_id
            PARAMS_DOCKER_ASR_PATH=${PARAMS_DOWNLOAD_MODEL_DIR}/${PARAMS_ASR_ID}
            PARAMS_DOCKER_ASR_DIR=$(dirname "${PARAMS_DOCKER_ASR_PATH}")
            PARAMS_LOCAL_ASR_PATH=${PARAMS_FUNASR_LOCAL_MODELS_DIR}/${PARAMS_ASR_ID}
            PARAMS_LOCAL_ASR_DIR=$(dirname "${PARAMS_LOCAL_ASR_PATH}")
        else
            PARAMS_ASR_ID=""
            PARAMS_LOCAL_ASR_PATH=$model_id
            if [ ! -d "$PARAMS_LOCAL_ASR_PATH" ]; then
                echo -e "  ${RED}${PARAMS_LOCAL_ASR_PATH} does not exist, please set again.${PLAIN}"
            else
                model_name=$(basename "${PARAMS_LOCAL_ASR_PATH}")
                PARAMS_LOCAL_ASR_DIR=$(dirname "${PARAMS_LOCAL_ASR_PATH}")
                middle=${PARAMS_LOCAL_ASR_DIR#*"${PARAMS_FUNASR_LOCAL_MODELS_DIR}"}
                PARAMS_DOCKER_ASR_DIR=$PARAMS_DOWNLOAD_MODEL_DIR
                PARAMS_DOCKER_ASR_PATH=${PARAMS_DOCKER_ASR_DIR}/${middle}/${model_name}
            fi
        fi
    fi
    result=$(echo ${model_type} | grep "--vad_model")
    if [ "$result" != "" ]; then
        if [ $local_flag -eq 0 ]; then
            PARAMS_VAD_ID=$model_id
            PARAMS_DOCKER_VAD_PATH=${PARAMS_DOWNLOAD_MODEL_DIR}/${PARAMS_VAD_ID}
            PARAMS_DOCKER_VAD_DIR=$(dirname "${PARAMS_DOCKER_VAD_PATH}")
            PARAMS_LOCAL_VAD_PATH=${PARAMS_FUNASR_LOCAL_MODELS_DIR}/${PARAMS_VAD_ID}
            PARAMS_LOCAL_VAD_DIR=$(dirname "${PARAMS_LOCAL_VAD_PATH}")
        else
            PARAMS_VAD_ID=""
            PARAMS_LOCAL_VAD_PATH=$model_id
            if [ ! -d "$PARAMS_LOCAL_VAD_PATH" ]; then
                echo -e "  ${RED}${PARAMS_LOCAL_VAD_PATH} does not exist, please set again.${PLAIN}"
            else
                model_name=$(basename "${PARAMS_LOCAL_VAD_PATH}")
                PARAMS_LOCAL_VAD_DIR=$(dirname "${PARAMS_LOCAL_VAD_PATH}")
                middle=${PARAMS_LOCAL_VAD_DIR#*"${PARAMS_FUNASR_LOCAL_MODELS_DIR}"}
                PARAMS_DOCKER_VAD_DIR=$PARAMS_DOWNLOAD_MODEL_DIR
                PARAMS_DOCKER_VAD_PATH=${PARAMS_DOCKER_VAD_DIR}/${middle}/${model_name}
            fi
        fi
    fi
    result=$(echo $model_type | grep "--punc_model")
    if [ "$result" != "" ]; then
        if [ $local_flag -eq 0 ]; then
            PARAMS_PUNC_ID=$model_id
            PARAMS_DOCKER_PUNC_PATH=${PARAMS_DOWNLOAD_MODEL_DIR}/${PARAMS_PUNC_ID}
            PARAMS_DOCKER_PUNC_DIR=$(dirname "${PARAMS_DOCKER_PUNC_PATH}")
            PARAMS_LOCAL_PUNC_PATH=${PARAMS_FUNASR_LOCAL_MODELS_DIR}/${PARAMS_PUNC_ID}
            PARAMS_LOCAL_PUNC_DIR=$(dirname "${PARAMS_LOCAL_PUNC_PATH}")
        else
            model_name=$(basename "${PARAMS_LOCAL_PUNC_PATH}")
            PARAMS_LOCAL_PUNC_DIR=$(dirname "${PARAMS_LOCAL_PUNC_PATH}")
            middle=${PARAMS_LOCAL_PUNC_DIR#*"${PARAMS_FUNASR_LOCAL_MODELS_DIR}"}
            PARAMS_DOCKER_PUNC_DIR=$PARAMS_DOWNLOAD_MODEL_DIR
            PARAMS_DOCKER_PUNC_PATH=${PARAMS_DOCKER_PUNC_DIR}/${middle}/${model_name}
        fi
    fi
}

threadNumChange() {
    type=$1
    val=$2

    if [ -z "$val"]; then
        num=`expr ${val} + 0`
        if [ $num -ge 1 ] && [ $num -le 1024 ]; then
            result=$(echo ${type} | grep "--decode_thread_num")
            if [ "$result" != "" ]; then
                PARAMS_DECODER_THREAD_NUM=$num
            fi
            result=$(echo ${type} | grep "--io_thread_num")
            if [ "$result" != "" ]; then
                PARAMS_IO_THREAD_NUM=$num
            fi
        fi
    fi
}

portChange() {
    type=$1
    val=$2

    if [ ! -z "$val" ]; then
        port=`expr ${val} + 0`
        if [ $port -ge 1 ] && [ $port -le 65536 ]; then
            result=$(echo ${type} | grep "host_port")
            if [ "$result" != "" ]; then
                PARAMS_HOST_PORT=$port
            fi
            result=$(echo ${type} | grep "docker_port")
            if [ "$result" != "" ]; then
                PARAMS_DOCKER_PORT=$port
            fi
        fi
    fi
}

sampleClientRun(){
    echo -e "${YELLOW}Will download sample tools for the client to show how speech recognition works.${PLAIN}"

    download_cmd="curl ${DEFAULT_SAMPLES_URL} -o ${PARAMS_FUNASR_SAMPLES_LOCAL_PATH}"
    untar_cmd="tar -zxf ${PARAMS_FUNASR_SAMPLES_LOCAL_PATH} -C ${PARAMS_FUNASR_LOCAL_WORKSPACE}"

    if [ ! -f "$PARAMS_FUNASR_SAMPLES_LOCAL_PATH" ]; then
        $download_cmd
    fi    
    if [ -f "$PARAMS_FUNASR_SAMPLES_LOCAL_PATH" ]; then
        $untar_cmd
    fi
    if [ -d "$PARAMS_FUNASR_SAMPLES_LOCAL_DIR" ]; then
        echo -e "  Please select the client you want to run."
        menuSelection ${SAMPLE_CLIENTS[*]}
        result=$?
        index=`expr ${result} - 1`
        lang=${SAMPLE_CLIENTS[${index}]}
        echo

        server_ip="127.0.0.1"
        echo -e "  Please enter the IP of server, default(${CYAN}${server_ip}${PLAIN}): \c"
        read server_ip
        if [ -z "$server_ip" ]; then
            server_ip="127.0.0.1"
        fi

        host_port=`sed '/^PARAMS_HOST_PORT=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
        if [ -z "$host_port" ]; then
            host_port="10095"
        fi
        echo -e "  Please enter the port of server, default(${CYAN}${host_port}${PLAIN}): \c"
        read host_port
        if [ -z "$host_port" ]; then
            host_port=`sed '/^PARAMS_HOST_PORT=/!d;s/.*=//' ${DEFAULT_FUNASR_CONFIG_FILE}`
            if [ -z "$host_port" ]; then
                host_port="10095"
            fi
        fi

        wav_path="${PARAMS_FUNASR_SAMPLES_LOCAL_DIR}/audio/asr_example.wav"
        echo -e "  Please enter the audio path, default(${CYAN}${wav_path}${PLAIN}): \c"
        read WAV_PATH
        if [ -z "$wav_path" ]; then
            wav_path="${PARAMS_FUNASR_SAMPLES_LOCAL_DIR}/audio/asr_example.wav"
        fi

        echo
        pre_cmd=”“
        case "$lang" in
            Linux_Cpp)
                pre_cmd="export LD_LIBRARY_PATH=${PARAMS_FUNASR_SAMPLES_LOCAL_DIR}/cpp/libs:\$LD_LIBRARY_PATH"
                client_exec="${PARAMS_FUNASR_SAMPLES_LOCAL_DIR}/cpp/funasr-wss-client"
                run_cmd="${client_exec} --server-ip ${server_ip} --port ${host_port} --wav-path ${wav_path}"
                echo -e "  Run ${BLUE}${pre_cmd}${PLAIN}"
                $pre_cmd
                echo
                ;;
            Python)
                client_exec="${PARAMS_FUNASR_SAMPLES_LOCAL_DIR}/python/wss_client_asr.py"
                run_cmd="python3 ${client_exec} --host ${server_ip} --port ${host_port} --mode offline --audio_in ${wav_path} --send_without_sleep --output_dir ${PARAMS_FUNASR_SAMPLES_LOCAL_DIR}/python"
                pre_cmd="pip3 install click>=8.0.4"
                echo -e "  Run ${BLUE}${pre_cmd}${PLAIN}"
                $pre_cmd
                echo
                pre_cmd="pip3 install -r ${PARAMS_FUNASR_SAMPLES_LOCAL_DIR}/python/requirements_client.txt"
                echo -e "  Run ${BLUE}${pre_cmd}${PLAIN}"
                $pre_cmd
                echo
                ;;
            *)
                echo "${lang} is not supported."
                ;;
        esac

        echo -e "  Run ${BLUE}${run_cmd}${PLAIN}"
        $run_cmd
        echo
        echo -e "  If failed, you can try (${GREEN}${run_cmd}${PLAIN}) in your Shell."
        echo
    fi
}

paramsConfigure(){
    initConfiguration
    initParameters
    selectDockerImages
    result=$?
    result=`expr ${result} + 0`
    if [ ${result} -eq 50 ]; then
        return 50
    fi

    setupHostPort
    complementParameters
}

# Display Help info
displayHelp(){
    echo -e "${UNDERLINE}Usage${PLAIN}:"
    echo -e "  $0 [OPTIONAL FLAGS]"
    echo
    echo -e "funasr-runtime-deploy-offline-cpu-zh.sh - a Bash script to install&run FunASR docker."
    echo
    echo -e "${UNDERLINE}Options${PLAIN}:"
    echo -e "   ${BOLD}-i, install, --install${PLAIN}    Install and run FunASR docker."
    echo -e "                install [--workspace] <workspace in local>"
    echo -e "   ${BOLD}-s, start  , --start${PLAIN}      Run FunASR docker with configuration that has already been set."
    echo -e "   ${BOLD}-p, stop   , --stop${PLAIN}       Stop FunASR docker."
    echo -e "   ${BOLD}-r, restart, --restart${PLAIN}    Restart FunASR docker."
    echo -e "   ${BOLD}-u, update , --update${PLAIN}     Update parameters that has already been set."
    echo -e "                update [--workspace] <workspace in local>"
    echo -e "                update [--asr_model | --vad_model | --punc_model] <model_id or local model path>"
    echo -e "                update [--host_port | --docker_port] <port number>"
    echo -e "                update [--decode_thread_num | io_thread_num] <the number of threads>"
    echo -e "   ${BOLD}-c, client , --client${PLAIN}     Get a client example to show how to initiate speech recognition."
    echo -e "   ${BOLD}-o, show   , --show${PLAIN}       Displays all parameters that have been set."
    echo -e "   ${BOLD}-v, version, --version${PLAIN}    Display current script version."
    echo -e "   ${BOLD}-h, help   , --help${PLAIN}       Display this help."
    echo
    echo -e "   Version    : ${scriptVersion} "
    echo -e "   Modify Date: ${scriptDate}"
}

parseInput(){
    local menu
    menu=($(echo "$@"))
    len=${#menu[@]}

    stage=""
    if [ $len -ge 2 ]; then
        for val in ${menu[@]}
        do
            result=$(echo $val | grep "\-\-")
            if [ "$result" != "" ]; then
                stage=$result
            else
                if [ "$stage" = "--workspace" ]; then
                    relativePathToFullPath $val
                    PARAMS_FUNASR_LOCAL_WORKSPACE=$full_path
                    if [ ! -z "$PARAMS_FUNASR_LOCAL_WORKSPACE" ]; then
                        mkdir -p $PARAMS_FUNASR_LOCAL_WORKSPACE
                    fi
                fi
            fi
        done
    fi
}

# OS
OSID=$(grep ^ID= /etc/os-release | cut -d= -f2)
OSVER=$(lsb_release -cs)
OSNUM=$(grep -oE  "[0-9.]+" /etc/issue)
CPUNUM=$(cat /proc/cpuinfo |grep "processor"|wc -l)
DOCKERINFO=$(sudo docker info | wc -l)
DOCKERINFOLEN=`expr ${DOCKERINFO} + 0`

# PARAMS
#  The workspace for FunASR in local
PARAMS_FUNASR_LOCAL_WORKSPACE=$DEFAULT_FUNASR_LOCAL_WORKSPACE
#  The dir stored sample code in local
PARAMS_FUNASR_SAMPLES_LOCAL_DIR=${PARAMS_FUNASR_LOCAL_WORKSPACE}/${DEFAULT_SAMPLES_DIR}
#  The path of sample code in local
PARAMS_FUNASR_SAMPLES_LOCAL_PATH=${PARAMS_FUNASR_LOCAL_WORKSPACE}/${DEFAULT_SAMPLES_NAME}.tar.gz
#  The dir stored models in local
PARAMS_FUNASR_LOCAL_MODELS_DIR="${PARAMS_FUNASR_LOCAL_WORKSPACE}/models"
#  The path of configuration in local
PARAMS_FUNASR_CONFIG_PATH="${PARAMS_FUNASR_LOCAL_WORKSPACE}/config"

#  The server excutor in local
PARAMS_DOCKER_EXEC_PATH=$DEFAULT_DOCKER_EXEC_PATH
#  The dir stored server excutor in docker
PARAMS_DOCKER_EXEC_DIR=$DEFAULT_DOCKER_EXEC_DIR

#  The dir for downloading model in docker
PARAMS_DOWNLOAD_MODEL_DIR=$DEFAULT_FUNASR_WORKSPACE_DIR
#  The Docker image name
PARAMS_DOCKER_IMAGE=""

#  The dir stored punc model in local
PARAMS_LOCAL_PUNC_DIR=""
#  The path of punc model in local
PARAMS_LOCAL_PUNC_PATH=""
#  The dir stored punc model in docker
PARAMS_DOCKER_PUNC_DIR=""
#  The path of punc model in docker
PARAMS_DOCKER_PUNC_PATH=""
#  The punc model ID in ModelScope
PARAMS_PUNC_ID="damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx"

#  The dir stored vad model in local
PARAMS_LOCAL_VAD_DIR=""
#  The path of vad model in local
PARAMS_LOCAL_VAD_PATH=""
#  The dir stored vad model in docker
PARAMS_DOCKER_VAD_DIR=""
#  The path of vad model in docker
PARAMS_DOCKER_VAD_PATH=""
#  The vad model ID in ModelScope
PARAMS_VAD_ID="damo/speech_fsmn_vad_zh-cn-16k-common-onnx"

#  The dir stored asr model in local
PARAMS_LOCAL_ASR_DIR=""
#  The path of asr model in local
PARAMS_LOCAL_ASR_PATH=""
#  The dir stored asr model in docker
PARAMS_DOCKER_ASR_DIR=""
#  The path of asr model in docker
PARAMS_DOCKER_ASR_PATH=""
#  The asr model ID in ModelScope
PARAMS_ASR_ID="damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-onnx"

PARAMS_HOST_PORT="10095"
PARAMS_DOCKER_PORT="10095"
PARAMS_DECODER_THREAD_NUM="32"
PARAMS_IO_THREAD_NUM="8"


echo -e "#############################################################"
echo -e "#          ${RED}OS${PLAIN}: ${OSID} ${OSNUM} ${OSVER}"
echo -e "#      ${RED}Kernel${PLAIN}: $(uname -m) Linux $(uname -r)"
echo -e "#         ${RED}CPU${PLAIN}: $(grep 'model name' /proc/cpuinfo | uniq | awk -F : '{print $2}' | sed 's/^[ \t]*//g' | sed 's/ \+/ /g') "
echo -e "#     ${RED}CPU NUM${PLAIN}: ${CPUNUM}"
echo -e "#         ${RED}RAM${PLAIN}: $(cat /proc/meminfo | grep 'MemTotal' | awk -F : '{print $2}' | sed 's/^[ \t]*//g') "
echo -e "#"
echo -e "#     ${RED}Version${PLAIN}: ${scriptVersion} "
echo -e "# ${RED}Modify Date${PLAIN}: ${scriptDate}"
echo -e "#############################################################"
echo

# Initialization step
case "$1" in
    install|-i|--install)
        rootNess
        paramsFromDefault
        parseInput $@
        paramsConfigure
        result=$?
        result=`expr ${result} + 0`
        if [ ${result} -ne 50 ]; then
            showAllParams
            installFunasrDocker
            dockerRun
            result=$?
            stage=`expr ${result} + 0`
            if [ $stage -eq 98 ]; then
                dockerExit
                dockerRun
            fi
        fi
        ;;
    start|-s|--start)
        rootNess
        paramsFromDefault
        showAllParams
        dockerRun
        result=$?
        stage=`expr ${result} + 0`
        if [ $stage -eq 98 ]; then
            dockerExit
            dockerRun
        fi
        ;;
    restart|-r|--restart)
        rootNess
        paramsFromDefault
        showAllParams
        dockerExit
        dockerRun
        result=$?
        stage=`expr ${result} + 0`
        if [ $stage -eq 98 ]; then
            dockerExit
            dockerRun
        fi
        ;;
    stop|-p|--stop)
        rootNess
        paramsFromDefault
        dockerExit
        ;;
    update|-u|--update)
        rootNess
        paramsFromDefault

        if [ $# -eq 3 ]; then
            type=$2
            val=$3
            if [ "$type" = "--asr_model" ] || [ "$type" = "--vad_model" ] || [ "$type" = "--punc_model" ]; then
                modelChange $type $val
            elif [ "$type" = "--decode_thread_num" ] || [ "$type" = "--io_thread_num" ]; then
                threadNumChange $type $val
            elif [ "$type" = "--host_port" ] || [ "$type" = "--docker_port" ]; then
                portChange $type $val
            elif [ "$type" = "--workspace" ]; then
                relativePathToFullPath $val
                PARAMS_FUNASR_LOCAL_WORKSPACE=$full_path
                if [ ! -z "$PARAMS_FUNASR_LOCAL_WORKSPACE" ]; then
                    mkdir -p $PARAMS_FUNASR_LOCAL_WORKSPACE
                fi
            else
                displayHelp
            fi
        else
            displayHelp
        fi

        initParameters
        complementParameters
        showAllParams
        dockerExit
        dockerRun
        result=$?
        stage=`expr ${result} + 0`
        if [ $stage -eq 98 ]; then
            dockerExit
            dockerRun
        fi
        ;;
    client|-c|--client)
        rootNess
        paramsFromDefault
        parseInput $@
        sampleClientRun
        ;;
    show|-o|--show)
        rootNess
        paramsFromDefault
        showAllParams
        ;;
    *)
        displayHelp
        exit 0
        ;;
esac

 funasr/runtime/docs/SDK_advanced_guide_offline.md

New file
@@ -0,0 +1,259 @@
 # Advanced Development Guide (File transcription service)
 
FunASR provides a Chinese offline file transcription service that can be deployed locally or on a cloud server with just one click. The core of the service is the FunASR runtime SDK, which has been open-sourced. FunASR-runtime combines various capabilities such as speech endpoint detection (VAD), large-scale speech recognition (ASR) using Paraformer-large, and punctuation detection (PUNC), which have all been open-sourced by the speech laboratory of DAMO Academy on the Modelscope community. This enables accurate and efficient high-concurrency transcription of audio files.

This document serves as a development guide for the FunASR offline file transcription service. If you wish to quickly experience the offline file transcription service, please refer to the one-click deployment example for the FunASR offline file transcription service ([docs](./SDK_tutorial.md)).

## Installation of Docker

The following steps are for manually installing Docker and Docker images. If your Docker image has already been launched, you can ignore this step.

### Installation of Docker environment

```shell
# Ubuntu：
curl -fsSL https://test.docker.com -o test-docker.sh 
sudo sh test-docker.sh 
# Debian：
curl -fsSL https://get.docker.com -o get-docker.sh 
sudo sh get-docker.sh 
# CentOS：
curl -fsSL https://get.docker.com | bash -s docker --mirror Aliyun 
# MacOS：
brew install --cask --appdir=/Applications docker
```

More details could ref to [docs](https://alibaba-damo-academy.github.io/FunASR/en/installation/docker.html)

### Starting Docker

```shell
sudo systemctl start docker
```

### Pulling and launching images

Use the following command to pull and launch the Docker image for the FunASR runtime-SDK:

```shell
sudo docker pull registry.cn-hangzhou.aliyuncs.com/funasr_repo/funasr:funasr-runtime-sdk-cpu-latest

sudo docker run -p 10095:10095 -it --privileged=true -v /root:/workspace/models registry.cn-hangzhou.aliyuncs.com/funasr_repo/funasr:funasr-runtime-sdk-cpu-latest
```

Introduction to command parameters: 
```text
-p <host port>:<mapped docker port>: In the example, host machine (ECS) port 10095 is mapped to port 10095 in the Docker container. Make sure that port 10095 is open in the ECS security rules.

-v <host path>:<mounted Docker path>: In the example, the host machine path /root is mounted to the Docker path /workspace/models.

```


## Starting the server

Use the flollowing script to start the server ：
```shell
./run_server.sh --vad-dir damo/speech_fsmn_vad_zh-cn-16k-common-onnx \
  --model-dir damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-onnx  \
  --punc-dir damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx
```

More details about the script run_server.sh:

The FunASR-wss-server supports downloading models from Modelscope. You can set the model download address (--download-model-dir, default is /workspace/models) and the model ID (--model-dir, --vad-dir, --punc-dir). Here is an example:

```shell
cd /workspace/FunASR/funasr/runtime/websocket/build/bin
./funasr-wss-server  \
  --download-model-dir /workspace/models \
  --model-dir damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-onnx \
  --vad-dir damo/speech_fsmn_vad_zh-cn-16k-common-onnx \
  --punc-dir damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx \
  --decoder-thread-num 32 \
  --io-thread-num  8 \
  --port 10095 \
  --certfile  ../../../ssl_key/server.crt \
  --keyfile ../../../ssl_key/server.key
 ```

Introduction to command parameters: 

```text
--download-model-dir: Model download address, download models from Modelscope by setting the model ID.
--model-dir: Modelscope model ID.
--quantize: True for quantized ASR model, False for non-quantized ASR model. Default is True.
--vad-dir: Modelscope model ID.
--vad-quant: True for quantized VAD model, False for non-quantized VAD model. Default is True.
--punc-dir: Modelscope model ID.
--punc-quant: True for quantized PUNC model, False for non-quantized PUNC model. Default is True.
--port: Port number that the server listens on. Default is 10095.
--decoder-thread-num: Number of inference threads that the server starts. Default is 8.
--io-thread-num: Number of IO threads that the server starts. Default is 1.
--certfile <string>: SSL certificate file. Default is ../../../ssl_key/server.crt.
--keyfile <string>: SSL key file. Default is ../../../ssl_key/server.key.
```

The FunASR-wss-server also supports loading models from a local path (see Preparing Model Resources for detailed instructions on preparing local model resources). Here is an example:

```shell
cd /workspace/FunASR/funasr/runtime/websocket/build/bin
./funasr-wss-server  \
  --model-dir /workspace/models/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-onnx \
  --vad-dir /workspace/models/damo/speech_fsmn_vad_zh-cn-16k-common-onnx \
  --punc-dir /workspace/models/damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx \
  --decoder-thread-num 32 \
  --io-thread-num  8 \
  --port 10095 \
  --certfile  ../../../ssl_key/server.crt \
  --keyfile ../../../ssl_key/server.key
 ```


## Preparing Model Resources

If you choose to download models from Modelscope through the FunASR-wss-server, you can skip this step. The vad, asr, and punc model resources in the offline file transcription service of FunASR are all from Modelscope. The model addresses are shown in the table below:

| Model | Modelscope url                                                                                                   |
|-------|------------------------------------------------------------------------------------------------------------------|
| VAD   | https://www.modelscope.cn/models/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/summary |
| ASR   | https://www.modelscope.cn/models/damo/speech_fsmn_vad_zh-cn-16k-common-pytorch/summary                           |
| PUNC  | https://www.modelscope.cn/models/damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch/summary               |

The offline file transcription service deploys quantized ONNX models. Below are instructions on how to export ONNX models and their quantization. You can choose to export ONNX models from Modelscope, local files, or finetuned resources: 

### Exporting ONNX models from Modelscope

Download the corresponding model with the given model name from the Modelscope website, and then export the quantized ONNX model

```shell
python -m funasr.export.export_model \
--export-dir ./export \
--type onnx \
--quantize True \
--model-name damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch \
--model-name damo/speech_fsmn_vad_zh-cn-16k-common-pytorch \
--model-name damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch
```

Introduction to command parameters:

```text
--model-name: The name of the model on Modelscope, for example: damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch
--export-dir: The export directory of ONNX model.
--type: Model type, currently supports ONNX and torch.
--quantize: Quantize the int8 model.
```

### Exporting ONNX models from local files

Set the model name to the local path of the model, and export the quantized ONNX model:

```shell
python -m funasr.export.export_model --model-name /workspace/models/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch --export-dir ./export --type onnx --quantize True
```


### Exporting models from finetuned resources

If you want to deploy a finetuned model, you can follow these steps:
Rename the model you want to deploy after finetuning (for example, 10epoch.pb) to model.pb, and replace the original model.pb in Modelscope with this one. If the path of the replaced model is /path/to/finetune/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch, use the following command to convert the finetuned model to an ONNX model: 

```shell
python -m funasr.export.export_model --model-name /path/to/finetune/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch --export-dir ./export --type onnx --quantize True
```

## Starting the client

After completing the deployment of FunASR offline file transcription service on the server, you can test and use the service by following these steps. Currently, FunASR-bin supports multiple ways to start the client. The following are command-line examples based on python-client, c++-client, and custom client Websocket communication protocol: 

### python-client
```shell
python wss_client_asr.py --host "127.0.0.1" --port 10095 --mode offline --audio_in "./data/wav.scp" --send_without_sleep --output_dir "./results"
```

Introduction to command parameters:

```text
--host: the IP address of the server. It can be set to 127.0.0.1 for local testing.
--port: the port number of the server listener.
--audio_in: the audio input. Input can be a path to a wav file or a wav.scp file (a Kaldi-formatted wav list in which each line includes a wav_id followed by a tab and a wav_path).
--output_dir: the path to the recognition result output.
--ssl: whether to use SSL encryption. The default is to use SSL.
--mode: offline mode.
```

### c++-client
```shell
. /funasr-wss-client --server-ip 127.0.0.1 --port 10095 --wav-path test.wav --thread-num 1 --is-ssl 1
```

Introduction to command parameters:

```text
--host: the IP address of the server. It can be set to 127.0.0.1 for local testing.
--port: the port number of the server listener.
--audio_in: the audio input. Input can be a path to a wav file or a wav.scp file (a Kaldi-formatted wav list in which each line includes a wav_id followed by a tab and a wav_path).
--output_dir: the path to the recognition result output.
--ssl: whether to use SSL encryption. The default is to use SSL.
--mode: offline mode.
```

### Custom client

If you want to define your own client, the Websocket communication protocol is as follows:

```text
# First communication
{"mode": "offline", "wav_name": wav_name, "is_speaking": True}
# Send wav data
Bytes data
# Send end flag
{"is_speaking": False}
```

## How to customize service deployment

The code for FunASR-runtime is open source. If the server and client cannot fully meet your needs, you can further develop them based on your own requirements:

### C++ client

https://github.com/alibaba-damo-academy/FunASR/tree/main/funasr/runtime/websocket

### Python client

https://github.com/alibaba-damo-academy/FunASR/tree/main/funasr/runtime/python/websocket

### C++ server

#### VAD
```c++
// The use of the VAD model consists of two steps: FsmnVadInit and FsmnVadInfer:
FUNASR_HANDLE vad_hanlde=FsmnVadInit(model_path, thread_num);
// Where: model_path contains "model-dir" and "quantize", thread_num is the ONNX thread count;
FUNASR_RESULT result=FsmnVadInfer(vad_hanlde, wav_file.c_str(), NULL, 16000);
// Where: vad_hanlde is the return value of FunOfflineInit, wav_file is the path to the audio file, and sampling_rate is the sampling rate (default 16k).
```

See the usage example for details [docs](https://github.com/alibaba-damo-academy/FunASR/blob/main/funasr/runtime/onnxruntime/bin/funasr-onnx-offline-vad.cpp)

#### ASR
```text
// The use of the ASR model consists of two steps: FunOfflineInit and FunOfflineInfer:
FUNASR_HANDLE asr_hanlde=FunOfflineInit(model_path, thread_num);
// Where: model_path contains "model-dir" and "quantize", thread_num is the ONNX thread count;
FUNASR_RESULT result=FunOfflineInfer(asr_hanlde, wav_file.c_str(), RASR_NONE, NULL, 16000);
// Where: asr_hanlde is the return value of FunOfflineInit, wav_file is the path to the audio file, and sampling_rate is the sampling rate (default 16k).
```

See the usage example for details, [docs](https://github.com/alibaba-damo-academy/FunASR/blob/main/funasr/runtime/onnxruntime/bin/funasr-onnx-offline.cpp)

#### PUNC
```text
// The use of the PUNC model consists of two steps: CTTransformerInit and CTTransformerInfer:
FUNASR_HANDLE punc_hanlde=CTTransformerInit(model_path, thread_num);
// Where: model_path contains "model-dir" and "quantize", thread_num is the ONNX thread count;
FUNASR_RESULT result=CTTransformerInfer(punc_hanlde, txt_str.c_str(), RASR_NONE, NULL);
// Where: punc_hanlde is the return value of CTTransformerInit, txt_str is the text
```
See the usage example for details, [docs](https://github.com/alibaba-damo-academy/FunASR/blob/main/funasr/runtime/onnxruntime/bin/funasr-onnx-offline-punc.cpp)

 funasr/runtime/docs/SDK_advanced_guide_offline_zh.md

File was renamed from funasr/runtime/docs/SDK_advanced_guide_cn.md
@@ -35,9 +35,9 @@
通过下述命令拉取并启动FunASR runtime-SDK的docker镜像：

```shell
sudo docker pull registry.cn-hangzhou.aliyuncs.com/funasr_repo/funasr:funasr-runtime-sdk-cpu-0.0.1
sudo docker pull registry.cn-hangzhou.aliyuncs.com/funasr_repo/funasr:funasr-runtime-sdk-cpu-latest

sudo docker run -p 10095:10095 -it --privileged=true -v /root:/workspace/models registry.cn-hangzhou.aliyuncs.com/funasr_repo/funasr:funasr-runtime-sdk-cpu-0.0.1
sudo docker run -p 10095:10095 -it --privileged=true -v /root:/workspace/models registry.cn-hangzhou.aliyuncs.com/funasr_repo/funasr:funasr-runtime-sdk-cpu-latest
```

命令参数介绍：
@@ -52,6 +52,13 @@
## 服务端启动

docker启动之后，启动 funasr-wss-server服务程序：
```shell
./run_server.sh --vad-dir damo/speech_fsmn_vad_zh-cn-16k-common-onnx \
  --model-dir damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-onnx  \
  --punc-dir damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx
```

详细命令参数介绍：

funasr-wss-server支持从Modelscope下载模型，设置模型下载地址（--download-model-dir，默认为/workspace/models）及model ID（--model-dir、--vad-dir、--punc-dir）,示例如下：
```shell

 funasr/runtime/docs/SDK_tutorial.md

@@ -8,10 +8,14 @@

### Downloading Tools and Deployment

Run the following command to perform a one-click deployment of the FunASR runtime-SDK service. Follow the prompts to complete the deployment and running of the service. Currently, only Linux environments are supported, and for other environments, please refer to the Advanced SDK Development Guide. Due to network restrictions, the download of the funasr-runtime-deploy.sh one-click deployment tool may not proceed smoothly. If the tool has not been downloaded and entered into the one-click deployment tool after several seconds, please terminate it with Ctrl + C and run the following command again.
Run the following command to perform a one-click deployment of the FunASR runtime-SDK service. Follow the prompts to complete the deployment and running of the service. Currently, only Linux environments are supported, and for other environments, please refer to the Advanced SDK Development Guide ([docs](./SDK_advanced_guide_offline.md)). 

[//]: # (Due to network restrictions, the download of the funasr-runtime-deploy.sh one-click deployment tool may not proceed smoothly. If the tool has not been downloaded and entered into the one-click deployment tool after several seconds, please terminate it with Ctrl + C and run the following command again.)

```shell
curl -O https://raw.githubusercontent.com/alibaba-damo-academy/FunASR-APP/main/TransAudio/funasr-runtime-deploy.sh; sudo bash funasr-runtime-deploy.sh install
# For the users in China, you could install with the command:
# curl -O https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/shell/funasr-runtime-deploy.sh; sudo bash funasr-runtime-deploy.sh install
```

#### Details of Configuration

 funasr/runtime/docs/SDK_tutorial_cn.md

File was deleted

 funasr/runtime/docs/SDK_tutorial_zh.md

New file
@@ -0,0 +1,199 @@
# FunASR离线文件转写服务便捷部署教程

FunASR提供可便捷本地或者云端服务器部署的离线文件转写服务，内核为FunASR已开源runtime-SDK。
集成了达摩院语音实验室在Modelscope社区开源的语音端点检测(VAD)、Paraformer-large语音识别(ASR)、标点恢复(PUNC) 等相关能力，拥有完整的语音识别链路，可以将几十个小时的音频或视频识别成带标点的文字，而且支持上百路请求同时进行转写。

## 服务器配置

用户可以根据自己的业务需求，选择合适的服务器配置，推荐配置为：
- 配置1: （X86，计算型），4核vCPU，内存8G，单机可以支持大约32路的请求
- 配置2: （X86，计算型），16核vCPU，内存32G，单机可以支持大约64路的请求
- 配置3: （X86，计算型），64核vCPU，内存128G，单机可以支持大约200路的请求

详细性能测试报告（[点击此处](./benchmark_onnx_cpp.md)）

云服务厂商，针对新用户，有3个月免费试用活动，申请教程（[点击此处](./aliyun_server_tutorial.md)）

## 快速上手

### 服务端启动

下载部署工具`funasr-runtime-deploy-offline-cpu-zh.sh`

```shell
curl -O https://raw.githubusercontent.com/alibaba-damo-academy/FunASR/main/funasr/runtime/deploy_tools/funasr-runtime-deploy-offline-cpu-zh.sh;
# 如遇到网络问题，中国大陆用户，可以用个下面的命令：
# curl -O https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/shell/funasr-runtime-deploy-offline-cpu-zh.sh;
```

执行部署工具，在提示处输入回车键即可完成服务端安装与部署。目前便捷部署工具暂时仅支持Linux环境，其他环境部署参考开发指南（[点击此处](./SDK_advanced_guide_offline_zh.md)）
```shell
sudo bash funasr-runtime-deploy-offline-cpu-zh.sh install --workspace /root/funasr-runtime-resources
```

### 客户端测试与使用

运行上面安装指令后，会在/root/funasr-runtime-resources（默认安装目录）中下载客户端测试工具目录samples，
我们以Python语言客户端为例，进行说明，支持多种音频格式输入（.wav, .pcm, .mp3等），也支持视频输入(.mp4等)，以及多文件列表wav.scp输入，其他版本客户端请参考文档（[点击此处](#客户端用法详解)）

```shell
python3 wss_client_asr.py --host "127.0.0.1" --port 10095 --mode offline --audio_in "../audio/asr_example.wav" --output_dir "./results"
```

## 客户端用法详解

在服务器上完成FunASR服务部署以后，可以通过如下的步骤来测试和使用离线文件转写服务。
目前分别支持以下几种编程语言客户端

- [Python](#python-client)
- [CPP](#cpp-client)
- [html网页版本](#Html网页版)
- [Java](#Java-client)

更多版本客户端支持请参考[开发指南](./SDK_advanced_guide_offline_zh.md)

### python-client
若想直接运行client进行测试，可参考如下简易说明，以python版本为例：

```shell
python3 wss_client_asr.py --host "127.0.0.1" --port 10095 --mode offline --audio_in "../audio/asr_example.wav" --output_dir "./results"
```

命令参数说明：
```text
--host 为FunASR runtime-SDK服务部署机器ip，默认为本机ip（127.0.0.1），如果client与服务不在同一台服务器，需要改为部署机器ip
--port 10095 部署端口号
--mode offline表示离线文件转写
--audio_in 需要进行转写的音频文件，支持文件路径，文件列表wav.scp
--output_dir 识别结果保存路径
```

### cpp-client
进入samples/cpp目录后，可以用cpp进行测试，指令如下：
```shell
./funasr-wss-client --server-ip 127.0.0.1 --port 10095 --wav-path ../audio/asr_example.wav
```

命令参数说明：

```text
--server-ip 为FunASR runtime-SDK服务部署机器ip，默认为本机ip（127.0.0.1），如果client与服务不在同一台服务器，需要改为部署机器ip
--port 10095 部署端口号
--wav-path 需要进行转写的音频文件，支持文件路径
```

### Html网页版

在浏览器中打开 html/static/index.html，即可出现如下页面，支持麦克风输入与文件上传，直接进行体验

<img src="images/html.png"  width="900"/>

### Java-client

```shell
FunasrWsClient --host localhost --port 10095 --audio_in ./asr_example.wav --mode offline
```
详细可以参考文档（[点击此处](../java/readme.md)）

## 服务端用法详解

### 启动已经部署过的FunASR服务
一键部署后若出现重启电脑等关闭Docker的动作，可通过如下命令直接启动FunASR服务，启动配置为上次一键部署的设置。

```shell
sudo bash funasr-runtime-deploy-offline-cpu-zh.sh start
```

### 关闭FunASR服务

```shell
sudo bash funasr-runtime-deploy-offline-cpu-zh.sh stop
```

### 重启FunASR服务

根据上次一键部署的设置重启启动FunASR服务。
```shell
sudo bash funasr-runtime-deploy-offline-cpu-zh.sh restart
```

### 替换模型并重启FunASR服务

替换正在使用的模型，并重新启动FunASR服务。模型需为ModelScope中的ASR/VAD/PUNC模型，或者从ModelScope中模型finetune后的模型。

```shell
sudo bash funasr-runtime-deploy-offline-cpu-zh.sh update [--asr_model | --vad_model | --punc_model] <model_id or local model path>

e.g
sudo bash funasr-runtime-deploy-offline-cpu-zh.sh update --asr_model damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch
```

### 更新参数并重启FunASR服务

更新已配置参数，并重新启动FunASR服务生效。可更新参数包括宿主机和Docker的端口号，以及推理和IO的线程数量。

```shell
sudo bash funasr-runtime-deploy-offline-cpu-zh.sh update [--host_port | --docker_port] <port number>
sudo bash funasr-runtime-deploy-offline-cpu-zh.sh update [--decode_thread_num | --io_thread_num] <the number of threads>
sudo bash funasr-runtime-deploy-offline-cpu-zh.sh update [--workspace] <workspace in local>

e.g
sudo bash funasr-runtime-deploy-offline-cpu-zh.sh update --decode_thread_num 32
sudo bash funasr-runtime-deploy-offline-cpu-zh.sh update --workspace /root/funasr-runtime-resources
```


## 服务端启动过程配置详解

##### 选择FunASR Docker镜像
推荐选择1)使用我们的最新发布版镜像，也可选择历史版本。
```text
[1/5]
  Getting the list of docker images, please wait a few seconds.
    [DONE]

  Please choose the Docker image.
    1) registry.cn-hangzhou.aliyuncs.com/funasr_repo/funasr:funasr-runtime-sdk-cpu-0.1.0
  Enter your choice, default(1):
  You have chosen the Docker image: registry.cn-hangzhou.aliyuncs.com/funasr_repo/funasr:funasr-runtime-sdk-cpu-0.1.0
```


##### 设置宿主机提供给FunASR的端口
设置提供给Docker的宿主机端口，默认为10095。请保证此端口可用。
```text
[2/5]
  Please input the opened port in the host used for FunASR server.
  Setting the opened host port [1-65535], default(10095):
  The port of the host is 10095
  The port in Docker for FunASR server is 10095
```

## 联系我们

在您使用过程中，如果遇到问题，欢迎加入用户群进行反馈


|                                    钉钉用户群                                     |                                      微信               |
|:----------------------------------------------------------------------------:|:-----------------------------------------------------:|
| <div align="left"><img src="../../../docs/images/dingding.jpg" width="250"/> | <img src="../../../docs/images/wechat.png" width="232"/></div> |


## 视频demo

[点击此处]()
















 funasr/runtime/docs/aliyun_server_tutorial.md

@@ -2,16 +2,6 @@

我们以阿里云（[点此链接](https://www.aliyun.com/)）为例，演示如何申请云服务器

## 服务器配置

用户可以根据自己的业务需求，选择合适的服务器配置，推荐配置为：
- 配置一（高配）：X86架构，32/64核8369CPU，内存8G以上；
- 配置二：X86架构，32/64核8163CPU，内存8G以上；

详细性能测试报告：[点此链接](./benchmark_onnx_cpp.md)

我们以免费试用（1～3个月）为例，演示如何申请服务器流程，图文步骤如下：

### 登陆个人账号
打开阿里云官网[点此链接](https://www.aliyun.com/)，注册并登陆个人账号，如下图标号1所示

@@ -69,6 +59,6 @@

<img src="images/aliyun12.png"  width="900"/>

上图表示已经成功申请了云服务器，后续可以根据FunASR runtime-SDK部署文档进行一键部署（[点击此处]()）
上图表示已经成功申请了云服务器，后续可以根据FunASR runtime-SDK部署文档进行一键部署（[点击此处](./SDK_tutorial_cn.md)）



 funasr/runtime/docs/docker_offline_cpu_zh_lists

New file
@@ -0,0 +1,8 @@
DOCKER:
  funasr-runtime-sdk-cpu-0.1.0
DEFAULT_ASR_MODEL:
  damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-onnx
DEFAULT_VAD_MODEL:
  damo/speech_fsmn_vad_zh-cn-16k-common-onnx
DEFAULT_PUNC_MODEL:
  damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx

 funasr/runtime/docs/images/html.png

 funasr/runtime/python/websocket/README.md

@@ -36,7 +36,7 @@
```
##### Usage examples
```shell
python wss_srv_asr.py --port 10095 --asr_model "damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch"  --asr_model_online "damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online"
python wss_srv_asr.py --port 10095
```

## For the client
@@ -59,9 +59,8 @@
--words_max_print [max number of words to print] \
--audio_in [if set, loadding from wav.scp, else recording from mircrophone] \
--output_dir [if set, write the results to output_dir] \
--send_without_sleep [only set for offline] \
--ssl [1 for wss connect, 0 for ws, default is 1] \
--mode [`online` for streaming asr, `offline` for non-streaming, `2pass` for unifying streaming and non-streaming asr] \
--thread_num [thread_num for send data]
```

#### Usage examples
@@ -69,19 +68,19 @@
Recording from mircrophone
```shell
# --chunk_interval, "10": 600/10=60ms, "5"=600/5=120ms, "20": 600/12=30ms
python wss_client_asr.py --host "0.0.0.0" --port 10095 --mode offline --chunk_interval 10 --words_max_print 100
python wss_client_asr.py --host "0.0.0.0" --port 10095 --mode offline
```
Loadding from wav.scp(kaldi style)
```shell
# --chunk_interval, "10": 600/10=60ms, "5"=600/5=120ms, "20": 600/12=30ms
python wss_client_asr.py --host "0.0.0.0" --port 10095 --mode offline --chunk_interval 10 --words_max_print 100 --audio_in "./data/wav.scp" --send_without_sleep --output_dir "./results"
python wss_client_asr.py --host "0.0.0.0" --port 10095 --mode offline --audio_in "./data/wav.scp" --output_dir "./results"
```

##### ASR streaming client
Recording from mircrophone
```shell
# --chunk_size, "5,10,5"=600ms, "8,8,4"=480ms
python wss_client_asr.py --host "0.0.0.0" --port 10095 --mode online --chunk_size "5,10,5" --words_max_print 100
python wss_client_asr.py --host "0.0.0.0" --port 10095 --mode online --chunk_size "5,10,5"
```
Loadding from wav.scp(kaldi style)
```shell

 funasr/runtime/python/websocket/parse_args.py

File was deleted

 funasr/runtime/python/websocket/wss_client_asr.py

@@ -40,12 +40,12 @@
                    help="audio_in")
parser.add_argument("--send_without_sleep",
                    action="store_true",
                    default=False,
                    default=True,
                    help="if audio_in is set, send_without_sleep")
parser.add_argument("--test_thread_num",
parser.add_argument("--thread_num",
                    type=int,
                    default=1,
                    help="test_thread_num")
                    help="thread_num")
parser.add_argument("--words_max_print",
                    type=int,
                    default=10000,
@@ -161,7 +161,8 @@
                #voices.put(message)
                await websocket.send(message)
 
            sleep_duration = 0.001 if args.send_without_sleep else 60 * args.chunk_size[1] / args.chunk_interval / 1000
            sleep_duration = 0.001 if args.mode == "offline" else 60 * args.chunk_size[1] / args.chunk_interval / 1000
            
            await asyncio.sleep(sleep_duration)
    # when all data sent, we need to close websocket
    while not voices.empty():
@@ -175,9 +176,24 @@
         await asyncio.sleep(1)
    
    await websocket.close()
     
 
 


async def ws_send():
    global voices
    global websocket
    print("started to sending data!")
    while True:
        while not voices.empty():
            data = voices.get()
            voices.task_done()
            try:
                await websocket.send(data)
            except Exception as e:
                print('Exception occurred:', e)
                traceback.print_exc()
                exit(0)
            await asyncio.sleep(0.005)
        await asyncio.sleep(0.005)

 
             
@@ -261,9 +277,9 @@
            task = asyncio.create_task(record_from_scp(i, 1))
        else:
            task = asyncio.create_task(record_microphone())
        #task2 = asyncio.create_task(ws_send())
        task2 = asyncio.create_task(ws_send())
        task3 = asyncio.create_task(message(str(id)+"_"+str(i))) #processid+fileid
        await asyncio.gather(task, task3)
        await asyncio.gather(task, task2, task3)
  exit(0)
    

@@ -295,16 +311,16 @@
                    f'Not supported audio type: {audio_type}')

        total_len = len(wavs)
        if total_len >= args.test_thread_num:
            chunk_size = int(total_len / args.test_thread_num)
            remain_wavs = total_len - chunk_size * args.test_thread_num
        if total_len >= args.thread_num:
            chunk_size = int(total_len / args.thread_num)
            remain_wavs = total_len - chunk_size * args.thread_num
        else:
            chunk_size = 1
            remain_wavs = 0

        process_list = []
        chunk_begin = 0
        for i in range(args.test_thread_num):
        for i in range(args.thread_num):
            now_chunk_size = chunk_size
            if remain_wavs > 0:
                now_chunk_size = chunk_size + 1

 funasr/runtime/python/websocket/wss_srv_asr.py

@@ -5,8 +5,8 @@
import logging
import tracemalloc
import numpy as np
import argparse
import ssl
from parse_args import args
from modelscope.pipelines import pipeline
from modelscope.utils.constant import Tasks
from modelscope.utils.logger import get_logger
@@ -17,6 +17,54 @@
logger = get_logger(log_level=logging.CRITICAL)
logger.setLevel(logging.CRITICAL)

parser = argparse.ArgumentParser()
parser.add_argument("--host",
                    type=str,
                    default="0.0.0.0",
                    required=False,
                    help="host ip, localhost, 0.0.0.0")
parser.add_argument("--port",
                    type=int,
                    default=10095,
                    required=False,
                    help="grpc server port")
parser.add_argument("--asr_model",
                    type=str,
                    default="damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch",
                    help="model from modelscope")
parser.add_argument("--asr_model_online",
                    type=str,
                    default="damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online",
                    help="model from modelscope")
parser.add_argument("--vad_model",
                    type=str,
                    default="damo/speech_fsmn_vad_zh-cn-16k-common-pytorch",
                    help="model from modelscope")
parser.add_argument("--punc_model",
                    type=str,
                    default="damo/punc_ct-transformer_zh-cn-common-vad_realtime-vocab272727",
                    help="model from modelscope")
parser.add_argument("--ngpu",
                    type=int,
                    default=1,
                    help="0 for cpu, 1 for gpu")
parser.add_argument("--ncpu",
                    type=int,
                    default=4,
                    help="cpu cores")
parser.add_argument("--certfile",
                    type=str,
                    default="./ssl_key/server.crt",
                    required=False,
                    help="certfile for ssl")

parser.add_argument("--keyfile",
                    type=str,
                    default="./ssl_key/server.key",
                    required=False,
                    help="keyfile for ssl")
args = parser.parse_args()


websocket_users = set()


 funasr/runtime/readme.md

@@ -27,4 +27,4 @@

### Advanced Development Guide

The documentation mainly targets advanced developers who require modifications and customization of the service. It supports downloading model deployments from modelscope and also supports deploying models that users have fine-tuned. For detailed information, please refer to the documentation available by [docs](websocket/readme.md)
The documentation mainly targets advanced developers who require modifications and customization of the service. It supports downloading model deployments from modelscope and also supports deploying models that users have fine-tuned. For detailed information, please refer to the documentation available by [docs](./docs/SDK_advanced_guide_offline.md)

 funasr/runtime/readme_cn.md

@@ -9,23 +9,23 @@
- 中文离线文件转写服务（GPU版本），进行中
- 英文离线转写服务，进行中
- 流式语音识别服务，进行中
- 。。。
- 更多支持中


## 中文离线文件转写服务部署（CPU版本）

目前FunASR runtime-SDK-0.0.1版本已支持中文语音离线文件服务部署（CPU版本），拥有完整的语音识别链路，可以将几十个小时的音频识别成带标点的文字，而且支持上百路并发同时进行识别。

为了支持不同用户的需求，我们分别针对小白与高阶开发者，准备了不同的图文教程：

### 技术原理揭秘

文档介绍了背后技术原理，识别准确率，计算效率等，以及核心优势介绍：便捷、高精度、高效率、长音频链路，详细文档参考（[点击此处](https://mp.weixin.qq.com/s?__biz=MzA3MTQ0NTUyMw==&tempkey=MTIyNF84d05USjMxSEpPdk5GZXBJUFNJNzY0bU1DTkxhV19mcWY4MTNWQTJSYXhUaFgxOWFHZTZKR0JzWC1JRmRCdUxCX2NoQXg0TzFpNmVJX2R1WjdrcC02N2FEcUc3MDhzVVhpNWQ5clU4QUdqNFdkdjFYb18xRjlZMmc5c3RDOTl0U0NiRkJLb05ZZ0RmRlVkVjFCZnpXNWFBVlRhbXVtdWs4bUMwSHZnfn4%3D&chksm=1f2c3254285bbb42bc8f76a82e9c5211518a0bb1ff8c357d085c1b78f675ef2311f3be6e282c#rd)）
中文语音离线文件服务部署（CPU版本），拥有完整的语音识别链路，可以将几十个小时的长音频与视频识别成带标点的文字，而且支持上百路请求同时进行转写。
为了支持不同用户的需求，针对不同场景，准备了不同的图文教程：

### 便捷部署教程

文档主要针对小白用户与初级开发者，没有修改、定制需求，支持从modelscope中下载模型部署，也支持用户finetune后的模型部署，详细教程参考（[点击此处](./docs/SDK_tutorial_cn.md)）
适用场景为，对服务部署SDK无修改需求，部署模型来自于ModelScope，或者用户finetune，详细教程参考（[点击此处](./docs/SDK_tutorial_zh.md)）

### 高阶开发指南

文档主要针对高阶开发者，需要对服务进行修改与定制，支持从modelscope中下载模型部署，也支持用户finetune后的模型部署，详细文档参考（[点击此处](./docs/SDK_advanced_guide_cn.md)）
### 开发指南

适用场景为，对服务部署SDK有修改需求，部署模型来自于ModelScope，或者用户finetune，详细文档参考（[点击此处](./docs/SDK_advanced_guide_offline_zh.md)）

### 技术原理揭秘

文档介绍了背后技术原理，识别准确率，计算效率等，以及核心优势介绍：便捷、高精度、高效率、长音频链路，详细文档参考（[点击此处](https://mp.weixin.qq.com/s/DHQwbgdBWcda0w_L60iUww)）

 funasr/runtime/run_server.sh

@@ -1,4 +1,3 @@
#!/usr/bin/env bash

download_model_dir="/workspace/models"
model_dir="damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-onnx"
@@ -10,7 +9,7 @@
certfile="../../../ssl_key/server.crt"
keyfile="../../../ssl_key/server.key"

. ../../../egs/aishell/transformer/utils/parse_options.sh || exit 1;
. ../../egs/aishell/transformer/utils/parse_options.sh || exit 1;

cd /workspace/FunASR/funasr/runtime/websocket/build/bin
./funasr-wss-server  \
@@ -22,4 +21,5 @@
  --io-thread-num  ${io_thread_num} \
  --port ${port} \
  --certfile  ${certfile} \
  --keyfile ${keyfile}
  --keyfile ${keyfile}


 funasr/tasks/asr.py

@@ -47,6 +47,7 @@
from funasr.models.e2e_sa_asr import SAASRModel
from funasr.models.e2e_uni_asr import UniASR
from funasr.models.e2e_asr_transducer import TransducerModel, UnifiedTransducerModel
from funasr.models.e2e_asr_bat import BATModel
from funasr.models.encoder.abs_encoder import AbsEncoder
from funasr.models.encoder.conformer_encoder import ConformerEncoder, ConformerChunkEncoder
from funasr.models.encoder.data2vec_encoder import Data2VecEncoder
@@ -66,7 +67,7 @@
from funasr.models.postencoder.hugging_face_transformers_postencoder import (
    HuggingFaceTransformersPostEncoder,  # noqa: H301
)
from funasr.models.predictor.cif import CifPredictor, CifPredictorV2, CifPredictorV3
from funasr.models.predictor.cif import CifPredictor, CifPredictorV2, CifPredictorV3, BATPredictor
from funasr.models.preencoder.abs_preencoder import AbsPreEncoder
from funasr.models.preencoder.linear import LinearProjection
from funasr.models.preencoder.sinc import LightweightSincConvs
@@ -135,6 +136,7 @@
        timestamp_prediction=TimestampPredictor,
        rnnt=TransducerModel,
        rnnt_unified=UnifiedTransducerModel,
        bat=BATModel,
        sa_asr=SAASRModel,
    ),
    type_check=FunASRModel,
@@ -266,6 +268,7 @@
        ctc_predictor=None,
        cif_predictor_v2=CifPredictorV2,
        cif_predictor_v3=CifPredictorV3,
        bat_predictor=BATPredictor,
    ),
    type_check=None,
    default="cif_predictor",
@@ -1508,6 +1511,139 @@

        return model

class ASRBATTask(ASRTask):
    """ASR Boundary Aware Transducer Task definition."""

    num_optimizers: int = 1

    class_choices_list = [
        model_choices,
        frontend_choices,
        specaug_choices,
        normalize_choices,
        encoder_choices,
        rnnt_decoder_choices,
        joint_network_choices,
        predictor_choices,
    ]

    trainer = Trainer

    @classmethod
    def build_model(cls, args: argparse.Namespace) -> BATModel:
        """Required data depending on task mode.
        Args:
            cls: ASRBATTask object.
            args: Task arguments.
        Return:
            model: ASR BAT model.
        """
        assert check_argument_types()

        if isinstance(args.token_list, str):
            with open(args.token_list, encoding="utf-8") as f:
                token_list = [line.rstrip() for line in f]

            # Overwriting token_list to keep it as "portable".
            args.token_list = list(token_list)
        elif isinstance(args.token_list, (tuple, list)):
            token_list = list(args.token_list)
        else:
            raise RuntimeError("token_list must be str or list")
        vocab_size = len(token_list)
        logging.info(f"Vocabulary size: {vocab_size }")

        # 1. frontend
        if args.input_size is None:
            # Extract features in the model
            frontend_class = frontend_choices.get_class(args.frontend)
            frontend = frontend_class(**args.frontend_conf)
            input_size = frontend.output_size()
        else:
            # Give features from data-loader
            frontend = None
            input_size = args.input_size

        # 2. Data augmentation for spectrogram
        if args.specaug is not None:
            specaug_class = specaug_choices.get_class(args.specaug)
            specaug = specaug_class(**args.specaug_conf)
        else:
            specaug = None

        # 3. Normalization layer
        if args.normalize is not None:
            normalize_class = normalize_choices.get_class(args.normalize)
            normalize = normalize_class(**args.normalize_conf)
        else:
            normalize = None

        # 4. Encoder
        if getattr(args, "encoder", None) is not None:
            encoder_class = encoder_choices.get_class(args.encoder)
            encoder = encoder_class(input_size, **args.encoder_conf)
        else:
            encoder = Encoder(input_size, **args.encoder_conf)
        encoder_output_size = encoder.output_size()

        # 5. Decoder
        rnnt_decoder_class = rnnt_decoder_choices.get_class(args.rnnt_decoder)
        decoder = rnnt_decoder_class(
            vocab_size,
            **args.rnnt_decoder_conf,
        )
        decoder_output_size = decoder.output_size

        if getattr(args, "decoder", None) is not None:
            att_decoder_class = decoder_choices.get_class(args.decoder)

            att_decoder = att_decoder_class(
                vocab_size=vocab_size,
                encoder_output_size=encoder_output_size,
                **args.decoder_conf,
            )
        else:
            att_decoder = None
        # 6. Joint Network
        joint_network = JointNetwork(
            vocab_size,
            encoder_output_size,
            decoder_output_size,
            **args.joint_network_conf,
        )

        predictor_class = predictor_choices.get_class(args.predictor)
        predictor = predictor_class(**args.predictor_conf)

        # 7. Build model
        try:
            model_class = model_choices.get_class(args.model)
        except AttributeError:
            model_class = model_choices.get_class("rnnt_unified")

        model = model_class(
            vocab_size=vocab_size,
            token_list=token_list,
            frontend=frontend,
            specaug=specaug,
            normalize=normalize,
            encoder=encoder,
            decoder=decoder,
            att_decoder=att_decoder,
            joint_network=joint_network,
            predictor=predictor,
            **args.model_conf,
        )
        # 8. Initialize model
        if args.init is not None:
            raise NotImplementedError(
                "Currently not supported.",
                "Initialization part will be reworked in a short future.",
            )

        #assert check_return_type(model)

        return model

class ASRTaskSAASR(ASRTask):
    # If you need more than one optimizers, change this value

 funasr/version.txt

@@ -1 +1 @@
0.6.7
0.6.8