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

Merge branch dzh_straming_test into dev_gzf_deepspeed

Browse Source 
Title: align others with dev_gzf_deepspeed 

本次代码评审主要涉及添加CTC模块、优化MP3编码处理、简化音素令牌处理和引入新库，旨在增强模型的语音识别与文本转语音功能，同时改进音频处理逻辑。
Link: https://code.alibaba-inc.com/zhifu.gzf/FunASR/codereview/18421381
This commit is contained in:
shixian.shi 2024-09-18 22:26:21 +08:00
commit 9edbcd5420
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/llm_asr/model.py
View File

@ -3223,7 +3223,7 @@ class LLMASRXvecSlotTTS(nn.Module):
token_list.append(cur_token)
feat_list.append(feat)
# we should return this data to web page for playing.
mp3_data = self.convert_wav_to_mp3(wav)
mp3_data = self.convert_wav_to_mp3(wav, is_last)
wav_list.append(wav)
mp3_list.append(mp3_data)
@ -3329,7 +3329,7 @@ class LLMASRXvecSlotTTS(nn.Module):
states["chunk_idx"] = chunk_idx
if format == "mp3":
if cur_token is not None:
wav = self.convert_wav_to_mp3(wav)
wav = self.convert_wav_to_mp3(wav, is_last)
return cur_token, feat, wav
def write_mel_wav(self, output_dir, feat, wav, mp3, key):