speech_recognition/FunASR
1
0
Fork 0
mirror of https://github.com/modelscope/FunASR synced 2025-09-15 14:48:36 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

paraformer streaming bugfix

Browse Source
This commit is contained in:
游雁 2024-03-18 23:59:09 +08:00
parent dbbe718fc9
commit cbe2ea7e07
1 changed files with 7 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
funasr/models/paraformer_streaming/model.py
View File

@ -532,11 +532,13 @@ class ParaformerStreaming(Paraformer):
kwargs["is_final"] = _is_final and i == n -1 kwargs["is_final"] = _is_final and i == n -1
audio_sample_i = audio_sample[i*chunk_stride_samples:(i+1)*chunk_stride_samples] audio_sample_i = audio_sample[i*chunk_stride_samples:(i+1)*chunk_stride_samples]
if kwargs["is_final"] and len(audio_sample_i) < 960: if kwargs["is_final"] and len(audio_sample_i) < 960:
continue cache["encoder"]["tail_chunk"] = True
speech = cache["encoder"]["feats"]
# extract fbank feats speech_lengths = torch.tensor([speech.shape[1]], dtype=torch.int64).to(speech.device)
speech, speech_lengths = extract_fbank([audio_sample_i], data_type=kwargs.get("data_type", "sound"), else:
frontend=frontend, cache=cache["frontend"], is_final=kwargs["is_final"]) # 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() time3 = time.perf_counter()
meta_data["extract_feat"] = f"{time3 - time2:0.3f}" 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 meta_data["batch_data_time"] = speech_lengths.sum().item() * frontend.frame_shift * frontend.lfr_n / 1000