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
4,791 Commits 25 Branches 4 Tags 136 MiB
68c770f67c
Commit Graph

4791 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
木守
68c770f67c update 2024-09-11 17:57:48 +08:00
木守
9e69c9fa6a Merge branch 'dev_gzf_deepspeed' of http://gitlab.alibaba-inc.com/zhifu.gzf/FunASR into dev_gzf_deepspeed 2024-09-11 17:34:37 +08:00
木守
8ce7dad057 update 2024-09-11 17:34:33 +08:00
游雁
eebe29719b speech2speech 2024-09-11 17:02:52 +08:00
游雁
bdd66d1865 Merge branch 'dev_gzf_deepspeed' of gitlab.alibaba-inc.com:zhifu.gzf/FunASR into dev_gzf_deepspeed 
merge
 2024-09-11 16:16:24 +08:00
游雁
d46c2df2e1 speech2speech 2024-09-11 16:15:35 +08:00
木守
a333073086 update 2024-09-11 16:04:19 +08:00
木守
08865c2be4 Merge branch 'dev_gzf_deepspeed' of http://gitlab.alibaba-inc.com/zhifu.gzf/FunASR into dev_gzf_deepspeed 2024-09-11 15:52:48 +08:00
木守
c0782e19a8 update 2024-09-11 15:52:44 +08:00
游雁
6c51cc2e0a speech2speech 2024-09-11 12:30:03 +08:00
zhifu.gzf
11dcc53e40 Merge branch dzh_tts_4o into dev_gzf_deepspeed 
Title: add time counter for FM 

本次代码评审主要涉及在多个文件中添加了时间记录（使用`time.time()`），引入了新的模型类`LLMASRXvecSlotTTS`，增强了文本到语音（TTS）模型的功能，包括处理外部提示（outside_prompt），优化了模型结构和推理流程，并在多处进行了代码结构和功能的调整，以支持更复杂的模型训练和推理过程，特别是加入了对模型性能的监控和不同模块的融合。
Link: https://code.alibaba-inc.com/zhifu.gzf/FunASR/codereview/18311489
 2024-09-10 16:34:12 +08:00
neo.dzh
069665ef83 feat: Resolve conflict, auto committed by CodeFlow 2024-09-10 16:28:02 +08:00
志浩
b0f4cdc8e6 add time counter for FM 2024-09-10 15:49:30 +08:00
志浩
15c13be8f2 batchified cfg inference for FM 2024-09-10 15:42:11 +08:00
志浩
a24ef81aad add offline inference for e2e tts model in GPT-4o 2024-09-10 15:27:33 +08:00
木守
b15b4ca20f update 2024-09-10 12:12:29 +08:00
木守
03446f19ec Merge branch 'dev_gzf_deepspeed' of http://gitlab.alibaba-inc.com/zhifu.gzf/FunASR into dev_gzf_deepspeed 2024-09-10 12:05:43 +08:00
木守
f0280afefc update 2024-09-10 12:03:14 +08:00
志浩
bb21822b7f add multi-lingual tiktoken for 25Hz models 2024-09-10 11:34:10 +08:00
木守
6e22735f65 chore: update model paths in websocket server 2024-09-09 20:36:39 +08:00
木守
bd4309ea3c refactor: adjust parameters for websocket server 2024-09-09 20:31:08 +08:00
zhifu.gzf
0941f8edad Merge branch dev_gzf_llm2 into dev_gzf_deepspeed 
Title: wss 

本次代码评审主要加入了时间戳打印，用于追踪WebSocket服务器中语音处理的关键步骤时间点，并新增了一个多轮对话处理的WebSocket服务器实现，包括语音合成和模型推理过程，提高了代码的可追溯性和诊断能力。
Link: https://code.alibaba-inc.com/zhifu.gzf/FunASR/codereview/18228062
 2024-09-04 15:25:11 +08:00
游雁
b10a1ab523 wss 2024-09-04 12:53:40 +08:00
木守
245ba8fd45 Merge branch 'dev_gzf_deepspeed' of http://gitlab.alibaba-inc.com/zhifu.gzf/FunASR into dev_gzf_deepspeed 2024-09-03 19:45:35 +08:00
木守
281ebc3202 streaming 2024-09-03 19:45:29 +08:00
游雁
ef02bf322e wss 2024-09-03 14:42:27 +08:00
木守
467056fc2f streaming 2024-09-03 14:08:05 +08:00
yangyexin.yyx
6aa4f30555 Merge branch llm_dev_gzf into dev_gzf_deepspeed 
Title: lora 

本次代码评审主要涉及对一个基于WebSocket的语音识别与处理系统的更新，包括添加日志打印、修改默认参数、增加对LoRA模型的支持、调整错误处理逻辑、优化音频处理和文本显示逻辑，以及代码结构和注释的若干改进，旨在提升系统稳定性和用户体验。
Link: https://code.alibaba-inc.com/zhifu.gzf/FunASR/codereview/18202582
 2024-09-03 13:55:59 +08:00
zhifu.gzf
963472437c feat: Resolve conflict, auto committed by CodeFlow 2024-09-03 13:54:42 +08:00
游雁
a9668ad075 lora 2024-09-03 12:51:16 +08:00
游雁
9f8107ff9e lora 2024-09-03 11:48:49 +08:00
游雁
9de6a04db9 lora 2024-09-03 11:34:07 +08:00
游雁
5a2d8a38ac lora 2024-09-03 11:26:25 +08:00
游雁
29717f4361 lora 2024-09-03 11:23:34 +08:00
游雁
8fb3ce8796 ws 2024-09-02 19:15:59 +08:00
木守
57729f40a6 streaming 2024-09-02 19:03:48 +08:00
木守
013f02e3a3 streaming 2024-09-02 15:49:30 +08:00
木守
561de8db92 streaming 2024-09-02 15:02:31 +08:00
木守
b60b9b6454 streaming 2024-09-02 14:27:09 +08:00
木守
b7f0e6894f streaming 2024-09-02 14:23:45 +08:00
zhifu.gzf
623fd16f34 Merge branch dev_lr_deepspeed into dev_gzf_deepspeed 
Title: Add llm tts to client process. 

本次代码评审主要增强了 WebSocket 服务端和客户端的语音合成（TTS）功能，添加了语音数据发送、接收处理和计数逻辑，优化了模块结构，引入了`NlsTtsSynthesizer`类来管理语音合成流程，并调整了错误处理和连接管理，使得语音传输更稳定且可追踪。
Link: https://code.alibaba-inc.com/zhifu.gzf/FunASR/codereview/18178626
 2024-09-02 10:22:55 +08:00
木守
32657bf651 streaming asr/s2tt 2024-08-29 18:53:24 +08:00
木守
47492abac5 streaming asr/s2tt 2024-08-29 16:05:08 +08:00
jichi.lr
e99201c9b0 Add llm tts to client process. 2024-08-29 11:42:20 +08:00
yangyexin.yyx
038f752e58 streaming 2024-08-28 14:29:04 +08:00
yangyexin.yyx
71b6ecbb39 streaming 2024-08-28 14:20:14 +08:00
yangyexin.yyx
1eb7507c24 Merge branch 'dev_gzf_deepspeed' of http://gitlab.alibaba-inc.com/zhifu.gzf/FunASR into dev_gzf_deepspeed 2024-08-28 09:48:41 +08:00
yangyexin.yyx
366603d4ed streaming asr 2024-08-28 09:48:06 +08:00
游雁
7674885f5e age gender 2024-08-27 16:35:28 +08:00
游雁
032d429a94 wss llm 2024-08-26 17:52:30 +08:00