export

游雁

2023-03-28 c776f8afc0c24dd687b14035b94808fd1a5d48bf

export

 funasr/export/export_model.py

@@ -161,31 +161,38 @@
    
    def export(self,
               tag_name: str = 'damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch',
               mode: str = 'paraformer',
               mode: str = None,
               ):
        
        model_dir = tag_name
        if model_dir.startswith('damo/'):
        if model_dir.startswith('damo'):
            from modelscope.hub.snapshot_download import snapshot_download
            model_dir = snapshot_download(model_dir, cache_dir=self.cache_dir)
        asr_train_config = os.path.join(model_dir, 'config.yaml')
        asr_model_file = os.path.join(model_dir, 'model.pb')
        cmvn_file = os.path.join(model_dir, 'am.mvn')
        json_file = os.path.join(model_dir, 'configuration.json')

        if mode is None:
            import json
            json_file = os.path.join(model_dir, 'configuration.json')
            with open(json_file, 'r') as f:
                config_data = json.load(f)
                mode = config_data['model']['model_config']['mode']
        if mode.startswith('paraformer'):
            from funasr.tasks.asr import ASRTaskParaformer as ASRTask
        elif mode.startswith('uniasr'):
            from funasr.tasks.asr import ASRTaskUniASR as ASRTask
            config = os.path.join(model_dir, 'config.yaml')
            model_file = os.path.join(model_dir, 'model.pb')
            cmvn_file = os.path.join(model_dir, 'am.mvn')
            model, asr_train_args = ASRTask.build_model_from_file(
                config, model_file, cmvn_file, 'cpu'
            )
            self.frontend = model.frontend
        elif mode.startswith('offline'):
            from funasr.tasks.vad import VADTask
            config = os.path.join(model_dir, 'vad.yaml')
            model_file = os.path.join(model_dir, 'vad.pb')
            cmvn_file = os.path.join(model_dir, 'vad.mvn')
            
        model, asr_train_args = ASRTask.build_model_from_file(
            asr_train_config, asr_model_file, cmvn_file, 'cpu'
        )
        self.frontend = model.frontend
            model, vad_infer_args = VADTask.build_model_from_file(
                config, model_file, 'cpu'
            )
        self._export(model, tag_name)
            


 funasr/export/models/__init__.py

@@ -1,13 +1,15 @@
from funasr.models.e2e_asr_paraformer import Paraformer, BiCifParaformer
from funasr.export.models.e2e_asr_paraformer import Paraformer as Paraformer_export
from funasr.export.models.e2e_asr_paraformer import BiCifParaformer as BiCifParaformer_export
from funasr.models.e2e_uni_asr import UniASR

from funasr.models.e2e_vad import E2EVadModel
from funasr.export.models.e2e_vad import E2EVadModel as E2EVadModel_export

def get_model(model, export_config=None):
    if isinstance(model, BiCifParaformer):
        return BiCifParaformer_export(model, **export_config)
    elif isinstance(model, Paraformer):
        return Paraformer_export(model, **export_config)
    elif isinstance(model, E2EVadModel):
        return E2EVadModel_export(model, **export_config)
    else:
        raise "Funasr does not support the given model type currently."

 funasr/export/models/e2e_vad.py

New file
@@ -0,0 +1,69 @@
from enum import Enum
from typing import List, Tuple, Dict, Any

import torch
from torch import nn
import math

from funasr.models.encoder.fsmn_encoder import FSMN
from funasr.export.models.encoder.fsmn_encoder import FSMN as FSMN_export

class E2EVadModel(nn.Module):
    def __init__(self, model,
                max_seq_len=512,
                feats_dim=560,
                model_name='model',
                **kwargs,):
        super(E2EVadModel, self).__init__()
        self.feats_dim = feats_dim
        self.max_seq_len = max_seq_len
        self.model_name = model_name
        if isinstance(model.encoder, FSMN):
            self.encoder = FSMN_export(model.encoder)
        else:
            raise "unsupported encoder"
        

    def forward(self, feats: torch.Tensor,
                       in_cache0: torch.Tensor,
                       in_cache1: torch.Tensor,
                       in_cache2: torch.Tensor,
                       in_cache3: torch.Tensor,
                       ):

        scores, cache0, cache1, cache2, cache3 = self.encoder(feats,
                                                              in_cache0,
                                                              in_cache1,
                                                              in_cache2,
                                                              in_cache3)  # return B * T * D
        return scores, cache0, cache1, cache2, cache3

    def get_dummy_inputs(self, frame=30):
        speech = torch.randn(1, frame, self.feats_dim)
        in_cache0 = torch.randn(1, 128, 19, 1)
        in_cache1 = torch.randn(1, 128, 19, 1)
        in_cache2 = torch.randn(1, 128, 19, 1)
        in_cache3 = torch.randn(1, 128, 19, 1)
        
        return (speech, in_cache0, in_cache1, in_cache2, in_cache3)

    # def get_dummy_inputs_txt(self, txt_file: str = "/mnt/workspace/data_fbank/0207/12345.wav.fea.txt"):
    #     import numpy as np
    #     fbank = np.loadtxt(txt_file)
    #     fbank_lengths = np.array([fbank.shape[0], ], dtype=np.int32)
    #     speech = torch.from_numpy(fbank[None, :, :].astype(np.float32))
    #     speech_lengths = torch.from_numpy(fbank_lengths.astype(np.int32))
    #     return (speech, speech_lengths)

    def get_input_names(self):
        return ['speech', 'in_cache0', 'in_cache1', 'in_cache2', 'in_cache3']

    def get_output_names(self):
        return ['logits', 'out_cache0', 'out_cache1', 'out_cache2', 'out_cache3']

    def get_dynamic_axes(self):
        return {
            'speech': {
                1: 'feats_length'
            },
        }

 funasr/export/models/encoder/fsmn_encoder.py

New file
@@ -0,0 +1,297 @@
from typing import Tuple, Dict
import copy

import numpy as np
import torch
import torch.nn as nn
import torch.nn.functional as F
from funasr.models.encoder.fsmn_encoder import BasicBlock

class LinearTransform(nn.Module):

    def __init__(self, input_dim, output_dim):
        super(LinearTransform, self).__init__()
        self.input_dim = input_dim
        self.output_dim = output_dim
        self.linear = nn.Linear(input_dim, output_dim, bias=False)

    def forward(self, input):
        output = self.linear(input)

        return output


class AffineTransform(nn.Module):

    def __init__(self, input_dim, output_dim):
        super(AffineTransform, self).__init__()
        self.input_dim = input_dim
        self.output_dim = output_dim
        self.linear = nn.Linear(input_dim, output_dim)

    def forward(self, input):
        output = self.linear(input)

        return output


class RectifiedLinear(nn.Module):

    def __init__(self, input_dim, output_dim):
        super(RectifiedLinear, self).__init__()
        self.dim = input_dim
        self.relu = nn.ReLU()
        self.dropout = nn.Dropout(0.1)

    def forward(self, input):
        out = self.relu(input)
        return out


class FSMNBlock(nn.Module):

    def __init__(
            self,
            input_dim: int,
            output_dim: int,
            lorder=None,
            rorder=None,
            lstride=1,
            rstride=1,
    ):
        super(FSMNBlock, self).__init__()

        self.dim = input_dim

        if lorder is None:
            return

        self.lorder = lorder
        self.rorder = rorder
        self.lstride = lstride
        self.rstride = rstride

        self.conv_left = nn.Conv2d(
            self.dim, self.dim, [lorder, 1], dilation=[lstride, 1], groups=self.dim, bias=False)

        if self.rorder > 0:
            self.conv_right = nn.Conv2d(
                self.dim, self.dim, [rorder, 1], dilation=[rstride, 1], groups=self.dim, bias=False)
        else:
            self.conv_right = None

    def forward(self, input: torch.Tensor, cache: torch.Tensor):
        x = torch.unsqueeze(input, 1)
        x_per = x.permute(0, 3, 2, 1)  # B D T C
        
        cache = cache.to(x_per.device)
        y_left = torch.cat((cache, x_per), dim=2)
        cache = y_left[:, :, -(self.lorder - 1) * self.lstride:, :]
        y_left = self.conv_left(y_left)
        out = x_per + y_left

        if self.conv_right is not None:
            # maybe need to check
            y_right = F.pad(x_per, [0, 0, 0, self.rorder * self.rstride])
            y_right = y_right[:, :, self.rstride:, :]
            y_right = self.conv_right(y_right)
            out += y_right

        out_per = out.permute(0, 3, 2, 1)
        output = out_per.squeeze(1)

        return output, cache


class BasicBlock_export(nn.Module):
    def __init__(self,
                 model,
                 ):
        super(BasicBlock_export, self).__init__()
        self.linear = model.linear
        self.fsmn_block = model.fsmn_block
        self.affine = model.affine
        self.relu = model.relu

    def forward(self, input: torch.Tensor, in_cache: torch.Tensor):
        x = self.linear(input)  # B T D
        # cache_layer_name = 'cache_layer_{}'.format(self.stack_layer)
        # if cache_layer_name not in in_cache:
        #     in_cache[cache_layer_name] = torch.zeros(x1.shape[0], x1.shape[-1], (self.lorder - 1) * self.lstride, 1)
        x, out_cache = self.fsmn_block(x, in_cache)
        x = self.affine(x)
        x = self.relu(x)
        return x, out_cache


# class FsmnStack(nn.Sequential):
#     def __init__(self, *args):
#         super(FsmnStack, self).__init__(*args)
#
#     def forward(self, input: torch.Tensor, in_cache: Dict[str, torch.Tensor]):
#         x = input
#         for module in self._modules.values():
#             x = module(x, in_cache)
#         return x


'''
FSMN net for keyword spotting
input_dim:              input dimension
linear_dim:             fsmn input dimensionll
proj_dim:               fsmn projection dimension
lorder:                 fsmn left order
rorder:                 fsmn right order
num_syn:                output dimension
fsmn_layers:            no. of sequential fsmn layers
'''


class FSMN(nn.Module):
    def __init__(
            self,
        model,
    ):
        super(FSMN, self).__init__()
        
        # self.input_dim = input_dim
        # self.input_affine_dim = input_affine_dim
        # self.fsmn_layers = fsmn_layers
        # self.linear_dim = linear_dim
        # self.proj_dim = proj_dim
        # self.output_affine_dim = output_affine_dim
        # self.output_dim = output_dim
        #
        # self.in_linear1 = AffineTransform(input_dim, input_affine_dim)
        # self.in_linear2 = AffineTransform(input_affine_dim, linear_dim)
        # self.relu = RectifiedLinear(linear_dim, linear_dim)
        # self.fsmn = FsmnStack(*[BasicBlock(linear_dim, proj_dim, lorder, rorder, lstride, rstride, i) for i in
        #                         range(fsmn_layers)])
        # self.out_linear1 = AffineTransform(linear_dim, output_affine_dim)
        # self.out_linear2 = AffineTransform(output_affine_dim, output_dim)
        # self.softmax = nn.Softmax(dim=-1)
        self.in_linear1 = model.in_linear1
        self.in_linear2 = model.in_linear2
        self.relu = model.relu
        # self.fsmn = model.fsmn
        self.out_linear1 = model.out_linear1
        self.out_linear2 = model.out_linear2
        self.softmax = model.softmax

        for i, d in enumerate(self.model.fsmn):
            if isinstance(d, BasicBlock):
                self.model.fsmn[i] = BasicBlock_export(d)

    def fuse_modules(self):
        pass

    def forward(
            self,
            input: torch.Tensor,
            *args,
    ):
        """
        Args:
            input (torch.Tensor): Input tensor (B, T, D)
            in_cache: when in_cache is not None, the forward is in streaming. The type of in_cache is a dict, egs,
            {'cache_layer_1': torch.Tensor(B, T1, D)}, T1 is equal to self.lorder. It is {} for the 1st frame
        """

        x = self.in_linear1(input)
        x = self.in_linear2(x)
        x = self.relu(x)
        # x4 = self.fsmn(x3, in_cache)  # self.in_cache will update automatically in self.fsmn
        out_caches = list()
        for i, d in enumerate(self.model.fsmn):
            in_cache = args[i]
            x, out_cache = d(x, in_cache)
            out_caches.append(out_cache)
        x = self.out_linear1(x)
        x = self.out_linear2(x)
        x = self.softmax(x)

        return x, *out_caches


'''
one deep fsmn layer
dimproj:                projection dimension, input and output dimension of memory blocks
dimlinear:              dimension of mapping layer
lorder:                 left order
rorder:                 right order
lstride:                left stride
rstride:                right stride
'''


class DFSMN(nn.Module):

    def __init__(self, dimproj=64, dimlinear=128, lorder=20, rorder=1, lstride=1, rstride=1):
        super(DFSMN, self).__init__()

        self.lorder = lorder
        self.rorder = rorder
        self.lstride = lstride
        self.rstride = rstride

        self.expand = AffineTransform(dimproj, dimlinear)
        self.shrink = LinearTransform(dimlinear, dimproj)

        self.conv_left = nn.Conv2d(
            dimproj, dimproj, [lorder, 1], dilation=[lstride, 1], groups=dimproj, bias=False)

        if rorder > 0:
            self.conv_right = nn.Conv2d(
                dimproj, dimproj, [rorder, 1], dilation=[rstride, 1], groups=dimproj, bias=False)
        else:
            self.conv_right = None

    def forward(self, input):
        f1 = F.relu(self.expand(input))
        p1 = self.shrink(f1)

        x = torch.unsqueeze(p1, 1)
        x_per = x.permute(0, 3, 2, 1)

        y_left = F.pad(x_per, [0, 0, (self.lorder - 1) * self.lstride, 0])

        if self.conv_right is not None:
            y_right = F.pad(x_per, [0, 0, 0, (self.rorder) * self.rstride])
            y_right = y_right[:, :, self.rstride:, :]
            out = x_per + self.conv_left(y_left) + self.conv_right(y_right)
        else:
            out = x_per + self.conv_left(y_left)

        out1 = out.permute(0, 3, 2, 1)
        output = input + out1.squeeze(1)

        return output


'''
build stacked dfsmn layers
'''


def buildDFSMNRepeats(linear_dim=128, proj_dim=64, lorder=20, rorder=1, fsmn_layers=6):
    repeats = [
        nn.Sequential(
            DFSMN(proj_dim, linear_dim, lorder, rorder, 1, 1))
        for i in range(fsmn_layers)
    ]

    return nn.Sequential(*repeats)


if __name__ == '__main__':
    fsmn = FSMN(400, 140, 4, 250, 128, 10, 2, 1, 1, 140, 2599)
    print(fsmn)

    num_params = sum(p.numel() for p in fsmn.parameters())
    print('the number of model params: {}'.format(num_params))
    x = torch.zeros(128, 200, 400)  # batch-size * time * dim
    y, _ = fsmn(x)  # batch-size * time * dim
    print('input shape: {}'.format(x.shape))
    print('output shape: {}'.format(y.shape))

    print(fsmn.to_kaldi_net())

 funasr/runtime/python/onnxruntime/funasr_onnx/punc_bin.py