From 9b4e9cc8a0311e5243d69b73ed073e7ea441982e Mon Sep 17 00:00:00 2001
From: 游雁 <zhifu.gzf@alibaba-inc.com>
Date: 星期三, 27 三月 2024 16:05:29 +0800
Subject: [PATCH] train update
---
funasr/models/paraformer_streaming/model.py | 147 ++++---------------------------------------------
1 files changed, 12 insertions(+), 135 deletions(-)
diff --git a/funasr/models/paraformer_streaming/model.py b/funasr/models/paraformer_streaming/model.py
index 63dba5d..499b487 100644
--- a/funasr/models/paraformer_streaming/model.py
+++ b/funasr/models/paraformer_streaming/model.py
@@ -531,10 +531,14 @@
for i in range(n):
kwargs["is_final"] = _is_final and i == n -1
audio_sample_i = audio_sample[i*chunk_stride_samples:(i+1)*chunk_stride_samples]
-
- # extract fbank feats
- speech, speech_lengths = extract_fbank([audio_sample_i], data_type=kwargs.get("data_type", "sound"),
- frontend=frontend, cache=cache["frontend"], is_final=kwargs["is_final"])
+ if kwargs["is_final"] and len(audio_sample_i) < 960:
+ cache["encoder"]["tail_chunk"] = True
+ speech = cache["encoder"]["feats"]
+ speech_lengths = torch.tensor([speech.shape[1]], dtype=torch.int64).to(speech.device)
+ else:
+ # extract fbank feats
+ speech, speech_lengths = extract_fbank([audio_sample_i], data_type=kwargs.get("data_type", "sound"),
+ frontend=frontend, cache=cache["frontend"], is_final=kwargs["is_final"])
time3 = time.perf_counter()
meta_data["extract_feat"] = f"{time3 - time2:0.3f}"
meta_data["batch_data_time"] = speech_lengths.sum().item() * frontend.frame_shift * frontend.lfr_n / 1000
@@ -560,135 +564,8 @@
ibest_writer["text"][key[0]] = text_postprocessed
return result, 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)
-
- if kwargs["decoder"] == "ParaformerSANMDecoder":
- kwargs["decoder"] = "ParaformerSANMDecoderOnline"
- decoder_class = tables.decoder_classes.get(kwargs["decoder"] + "Export")
- self.decoder = decoder_class(self.decoder, onnx=is_onnx)
-
- from funasr.utils.torch_function import MakePadMask
- from funasr.utils.torch_function import sequence_mask
-
- if is_onnx:
- self.make_pad_mask = MakePadMask(max_seq_len, flip=False)
- else:
- self.make_pad_mask = sequence_mask(max_seq_len, flip=False)
-
- self.forward = self._export_forward
-
- import copy
- import types
- encoder_model = copy.copy(self)
- decoder_model = copy.copy(self)
-
- # encoder
- encoder_model.forward = types.MethodType(ParaformerStreaming._export_encoder_forward, encoder_model)
- encoder_model.export_dummy_inputs = types.MethodType(ParaformerStreaming.export_encoder_dummy_inputs, encoder_model)
- encoder_model.export_input_names = types.MethodType(ParaformerStreaming.export_encoder_input_names, encoder_model)
- encoder_model.export_output_names = types.MethodType(ParaformerStreaming.export_encoder_output_names, encoder_model)
- encoder_model.export_dynamic_axes = types.MethodType(ParaformerStreaming.export_encoder_dynamic_axes, encoder_model)
- encoder_model.export_name = types.MethodType(ParaformerStreaming.export_encoder_name, encoder_model)
-
- # decoder
- decoder_model.forward = types.MethodType(ParaformerStreaming._export_decoder_forward, decoder_model)
- decoder_model.export_dummy_inputs = types.MethodType(ParaformerStreaming.export_decoder_dummy_inputs, decoder_model)
- decoder_model.export_input_names = types.MethodType(ParaformerStreaming.export_decoder_input_names, decoder_model)
- decoder_model.export_output_names = types.MethodType(ParaformerStreaming.export_decoder_output_names, decoder_model)
- decoder_model.export_dynamic_axes = types.MethodType(ParaformerStreaming.export_decoder_dynamic_axes, decoder_model)
- decoder_model.export_name = types.MethodType(ParaformerStreaming.export_decoder_name, decoder_model)
-
- return encoder_model, decoder_model
-
- def export_encoder_forward(
- self,
- speech: torch.Tensor,
- speech_lengths: torch.Tensor,
- ):
- # a. To device
- batch = {"speech": speech, "speech_lengths": speech_lengths, "online": True}
- # batch = to_device(batch, device=self.device)
-
- enc, enc_len = self.encoder(**batch)
- mask = self.make_pad_mask(enc_len)[:, None, :]
- alphas, _ = self.predictor.forward_cnn(enc, mask)
-
- return enc, enc_len, alphas
-
- def export_encoder_dummy_inputs(self):
- speech = torch.randn(2, 30, 560)
- speech_lengths = torch.tensor([6, 30], dtype=torch.int32)
- return (speech, speech_lengths)
-
- def export_encoder_input_names(self):
- return ['speech', 'speech_lengths']
-
- def export_encoder_output_names(self):
- return ['enc', 'enc_len', 'alphas']
-
- def export_encoder_dynamic_axes(self):
- return {
- 'speech': {
- 0: 'batch_size',
- 1: 'feats_length'
- },
- 'speech_lengths': {
- 0: 'batch_size',
- },
- 'enc': {
- 0: 'batch_size',
- 1: 'feats_length'
- },
- 'enc_len': {
- 0: 'batch_size',
- },
- 'alphas': {
- 0: 'batch_size',
- 1: 'feats_length'
- },
- }
-
- def export_encoder_name(self):
- return "model.onnx"
-
- def export_decoder_forward(
- self,
- enc: torch.Tensor,
- enc_len: torch.Tensor,
- acoustic_embeds: torch.Tensor,
- acoustic_embeds_len: torch.Tensor,
- *args,
- ):
- decoder_out, out_caches = self.decoder(enc, enc_len, acoustic_embeds, acoustic_embeds_len, *args)
- sample_ids = decoder_out.argmax(dim=-1)
-
- return decoder_out, sample_ids, out_caches
-
- def export_decoder_dummy_inputs(self):
- dummy_inputs = self.decoder.get_dummy_inputs(enc_size=self.encoder._output_size)
- return dummy_inputs
-
- def export_decoder_input_names(self):
-
- return self.decoder.get_input_names()
-
- def export_decoder_output_names(self):
-
- return self.decoder.get_output_names()
-
- def export_decoder_dynamic_axes(self):
- return self.decoder.get_dynamic_axes()
- def export_decoder_name(self):
- return "decoder.onnx"
\ No newline at end of file
+ def export(self, **kwargs):
+ from .export_meta import export_rebuild_model
+ models = export_rebuild_model(model=self, **kwargs)
+ return models
\ No newline at end of file
--
Gitblit v1.9.1