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

add extract token run_mode

Browse Source
This commit is contained in:
志浩 2024-09-24 17:15:30 +08:00
parent 1fb762d9be
commit ce5b79d234
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
funasr/models/sense_voice/model_small.py
View File

@ -2034,11 +2034,11 @@ class SenseVoiceL(nn.Module):
lfr_n = frontend.lfr_n if hasattr(frontend, "lfr_n") else 1
meta_data["batch_data_time"] = speech_lengths.sum().item() * frame_shift * lfr_n / 1000
speech = speech.to(device=kwargs["device"])[0, :, :]
speech = speech.to(device=kwargs["device"])
speech_lengths = speech_lengths.to(device=kwargs["device"])
(outs, ret_dict), out_lens = self.model.encoder(
speech.permute(0, 2, 1), speech_lengths,
speech, speech_lengths,
only_extract_tokens=True
)
tokens = ret_dict["indices"]