From 0d9384c8c0161259192cc3d676ca0d60e0d18e5c Mon Sep 17 00:00:00 2001
From: zhifu gao <zhifu.gzf@alibaba-inc.com>
Date: 星期一, 11 三月 2024 22:32:26 +0800
Subject: [PATCH] Dev gzf (#1474)
---
funasr/models/paraformer/model.py | 148 +++++++++++++++++++++++++++++++++++++------------
1 files changed, 111 insertions(+), 37 deletions(-)
diff --git a/funasr/models/paraformer/model.py b/funasr/models/paraformer/model.py
index 2cd9c88..41a1bf7 100644
--- a/funasr/models/paraformer/model.py
+++ b/funasr/models/paraformer/model.py
@@ -1,35 +1,30 @@
-import os
-import logging
-from typing import Union, Dict, List, Tuple, Optional
-
-import torch
-import torch.nn as nn
+#!/usr/bin/env python3
+# -*- encoding: utf-8 -*-
+# Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved.
+# MIT License (https://opensource.org/licenses/MIT)
import time
-
-from funasr.losses.label_smoothing_loss import (
- LabelSmoothingLoss, # noqa: H301
-)
-
-from funasr.models.paraformer.cif_predictor import mae_loss
-
-from funasr.models.transformer.utils.add_sos_eos import add_sos_eos
-from funasr.models.transformer.utils.nets_utils import make_pad_mask, pad_list
-from funasr.metrics.compute_acc import th_accuracy
-from funasr.train_utils.device_funcs import force_gatherable
-
-from funasr.models.paraformer.search import Hypothesis
-
+import torch
+import logging
from torch.cuda.amp import autocast
+from typing import Union, Dict, List, Tuple, Optional
-from funasr.utils.load_utils import load_audio_text_image_video, extract_fbank
-from funasr.utils import postprocess_utils
-from funasr.utils.datadir_writer import DatadirWriter
from funasr.register import tables
from funasr.models.ctc.ctc import CTC
+from funasr.utils import postprocess_utils
+from funasr.metrics.compute_acc import th_accuracy
+from funasr.utils.datadir_writer import DatadirWriter
+from funasr.models.paraformer.search import Hypothesis
+from funasr.models.paraformer.cif_predictor import mae_loss
+from funasr.train_utils.device_funcs import force_gatherable
+from funasr.losses.label_smoothing_loss import LabelSmoothingLoss
+from funasr.models.transformer.utils.add_sos_eos import add_sos_eos
+from funasr.models.transformer.utils.nets_utils import make_pad_mask, pad_list
+from funasr.utils.load_utils import load_audio_text_image_video, extract_fbank
+from funasr.train_utils.device_funcs import to_device
@tables.register("model_classes", "Paraformer")
-class Paraformer(nn.Module):
+class Paraformer(torch.nn.Module):
"""
Author: Speech Lab of DAMO Academy, Alibaba Group
Paraformer: Fast and Accurate Parallel Transformer for Non-autoregressive End-to-End Speech Recognition
@@ -38,7 +33,6 @@
def __init__(
self,
- # token_list: Union[Tuple[str, ...], List[str]],
specaug: Optional[str] = None,
specaug_conf: Optional[Dict] = None,
normalize: str = None,
@@ -160,8 +154,8 @@
self.predictor_bias = predictor_bias
self.sampling_ratio = sampling_ratio
self.criterion_pre = mae_loss(normalize_length=length_normalized_loss)
- # self.step_cur = 0
- #
+
+
self.share_embedding = share_embedding
if self.share_embedding:
self.decoder.embed = None
@@ -169,6 +163,7 @@
self.use_1st_decoder_loss = use_1st_decoder_loss
self.length_normalized_loss = length_normalized_loss
self.beam_search = None
+ self.error_calculator = None
def forward(
self,
@@ -439,7 +434,7 @@
# scorer.to(device=kwargs.get("device", "cpu"), dtype=getattr(torch, kwargs.get("dtype", "float32"))).eval()
self.beam_search = beam_search
- def generate(self,
+ def inference(self,
data_in,
data_lengths=None,
key: list=None,
@@ -456,11 +451,13 @@
self.nbest = kwargs.get("nbest", 1)
meta_data = {}
- if isinstance(data_in, torch.Tensor): # fbank
+ if isinstance(data_in, torch.Tensor) and kwargs.get("data_type", "sound") == "fbank": # fbank
speech, speech_lengths = data_in, data_lengths
if len(speech.shape) < 3:
speech = speech[None, :, :]
- if speech_lengths is None:
+ if speech_lengths is not None:
+ speech_lengths = speech_lengths.squeeze(-1)
+ else:
speech_lengths = speech.shape[1]
else:
# extract fbank feats
@@ -496,6 +493,8 @@
b, n, d = decoder_out.size()
if isinstance(key[0], (list, tuple)):
key = key[0]
+ if len(key) < b:
+ key = key*b
for i in range(b):
x = encoder_out[i, :encoder_out_lens[i], :]
am_scores = decoder_out[i, :pre_token_length[i], :]
@@ -517,9 +516,10 @@
nbest_hyps = [Hypothesis(yseq=yseq, score=score)]
for nbest_idx, hyp in enumerate(nbest_hyps):
ibest_writer = None
- if ibest_writer is None and kwargs.get("output_dir") is not None:
- writer = DatadirWriter(kwargs.get("output_dir"))
- ibest_writer = writer[f"{nbest_idx+1}best_recog"]
+ if kwargs.get("output_dir") is not None:
+ if not hasattr(self, "writer"):
+ self.writer = DatadirWriter(kwargs.get("output_dir"))
+ ibest_writer = self.writer[f"{nbest_idx+1}best_recog"]
# remove sos/eos and get results
last_pos = -1
if isinstance(hyp.yseq, list):
@@ -533,13 +533,12 @@
if tokenizer is not None:
# Change integer-ids to tokens
token = tokenizer.ids2tokens(token_int)
- text = tokenizer.tokens2text(token)
-
- text_postprocessed, _ = postprocess_utils.sentence_postprocess(token)
+ text_postprocessed = tokenizer.tokens2text(token)
+ if not hasattr(tokenizer, "bpemodel"):
+ text_postprocessed, _ = postprocess_utils.sentence_postprocess(token)
result_i = {"key": key[i], "text": text_postprocessed}
-
if ibest_writer is not None:
ibest_writer["token"][key[i]] = " ".join(token)
# ibest_writer["text"][key[i]] = text
@@ -550,3 +549,78 @@
return results, meta_data
+ def export(
+ self,
+ max_seq_len=512,
+ **kwargs,
+ ):
+ self.device = kwargs.get("device")
+ is_onnx = kwargs.get("type", "onnx") == "onnx"
+ encoder_class = tables.encoder_classes.get(kwargs["encoder"]+"Export")
+ self.encoder = encoder_class(self.encoder, onnx=is_onnx)
+
+ predictor_class = tables.predictor_classes.get(kwargs["predictor"]+"Export")
+ self.predictor = predictor_class(self.predictor, onnx=is_onnx)
+
+
+ decoder_class = tables.decoder_classes.get(kwargs["decoder"]+"Export")
+ self.decoder = decoder_class(self.decoder, onnx=is_onnx)
+
+ from funasr.utils.torch_function import sequence_mask
+
+
+ self.make_pad_mask = sequence_mask(max_seq_len, flip=False)
+
+ self.forward = self.export_forward
+
+ return self
+
+ def export_forward(
+ self,
+ speech: torch.Tensor,
+ speech_lengths: torch.Tensor,
+ ):
+ # a. To device
+ batch = {"speech": speech, "speech_lengths": speech_lengths}
+ batch = to_device(batch, device=self.device)
+
+ enc, enc_len = self.encoder(**batch)
+ mask = self.make_pad_mask(enc_len)[:, None, :]
+ pre_acoustic_embeds, pre_token_length, alphas, pre_peak_index = self.predictor(enc, mask)
+ pre_token_length = pre_token_length.floor().type(torch.int32)
+
+ decoder_out, _ = self.decoder(enc, enc_len, pre_acoustic_embeds, pre_token_length)
+ decoder_out = torch.log_softmax(decoder_out, dim=-1)
+ # sample_ids = decoder_out.argmax(dim=-1)
+
+ return decoder_out, pre_token_length
+
+ def export_dummy_inputs(self):
+ speech = torch.randn(2, 30, 560)
+ speech_lengths = torch.tensor([6, 30], dtype=torch.int32)
+ return (speech, speech_lengths)
+
+
+ def export_input_names(self):
+ return ['speech', 'speech_lengths']
+
+ def export_output_names(self):
+ return ['logits', 'token_num']
+
+ def export_dynamic_axes(self):
+ return {
+ 'speech': {
+ 0: 'batch_size',
+ 1: 'feats_length'
+ },
+ 'speech_lengths': {
+ 0: 'batch_size',
+ },
+ 'logits': {
+ 0: 'batch_size',
+ 1: 'logits_length'
+ },
+ }
+
+ def export_name(self, ):
+ return "model.onnx"
--
Gitblit v1.9.1