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 cross fade for 4o

Browse Source
This commit is contained in:
志浩 2024-09-18 15:47:50 +08:00
parent f3d43be322
commit 6e4362ac1f
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
funasr/models/llm_asr/tts_models/e2e_model.py
View File

@ -1057,7 +1057,7 @@ class UCTDXvecSlotModel(UpsampleCtcTokenDiffModel):
cur_token = cur_token[:, :cur_token.shape[1] - token_reback, :]
feat_reback = int(token_reback * self.fm_model.length_normalizer_ratio)
feat = feat[:, :, :feat.shape[2] - feat_reback]
wav_reback = int(1.0 / token_reback * wav_vocoder.sample_rate)
wav_reback = int(1.0 / 25 * wav_vocoder.sample_rate * token_reback)
wav = wav[:, :wav.shape[1] - wav_reback]
return cur_token, feat, wav, (prompt_token, token_reback), (prompt_audio, feat_reback)