227 Commits

Author	SHA1	Message	Date
lyuxiang.lx	9504c3f88b	fix flow matching training for zero shot inference	2024-08-01 10:54:27 +08:00
lyuxiang.lx	02f941d348	update model inference	2024-07-24 19:18:09 +08:00
lyuxiang.lx	a13411c561	add stream code	2024-07-23 00:02:30 +08:00
刘悦	f2939a9a50	Update file_utils.py ... fix flake8	2024-07-22 16:02:18 +08:00
刘悦	cf43100f66	Update file_utils.py ... add speed_change function	2024-07-22 11:44:39 +08:00
zhuyunfeng	6ae6ba3f77	Update frontend.py ... Fix the bug in handling anomalies for synthetic text ending with Chinese and English commas.	2024-07-14 22:33:56 +08:00
lyuxiang.lx	44aea805ea	add train cfg in flow matching	2024-07-11 17:36:59 +08:00
lyuxiang.lx	6cebcb3410	move use_spk_embedding to processor	2024-07-11 13:15:34 +08:00
lyuxiang.lx	0fd15bb12b	use spk_embedding when sft	2024-07-10 17:49:32 +08:00
lyuxiang.lx	a723ea375e	Merge branch 'main' of github.com:FunAudioLLM/CosyVoice into main	2024-07-10 16:42:11 +08:00
lyuxiang.lx	793a24862c	add constant lr scheduler	2024-07-10 16:37:25 +08:00
cyz	225b56de05	FIX: 修复自然语言控制生成音频时发生错误，异常信息如下：AttributeError: 'CosyVoiceFrontEnd' object has no attribute 'en_tn_model'	2024-07-10 12:02:41 +08:00
lyuxiang.lx	6a3e44242a	keep only embedding mean as spk embedding	2024-07-10 00:21:56 +08:00
lyuxiang.lx	ee9e87b4d3	add empty cache	2024-07-09 23:48:23 +08:00
lyuxiang.lx	7981796523	add WeTextProcessing	2024-07-09 23:37:54 +08:00
passerbya	69026d83bb	没有标点结尾时默认加上句号	2024-07-09 17:42:40 +08:00
passerbya	f9fe31f200	文本中没有标点时无法合成	2024-07-09 17:26:19 +08:00
passerbya	95b8866f3c	优先使用ttsfrd，ttsfrd不存在时使用WeTextProcessing	2024-07-09 17:25:55 +08:00
passerbya	88c8bf7b9e	更换前端为WeTextProcessing	2024-07-09 08:22:06 +08:00
passerbya	2f496104ec	半角句号会导致合成失败：RuntimeError: torch.cat(): expected a non-empty list of Tensors ... text='小明因为感冒,鼻子不通,讲话总带着齉音.' File "/usr/local/data/CosyVoice/cosyvoice/cli/cosyvoice.py", line 62, in inference_zero_shot return {'tts_speech': torch.concat(tts_speeches, dim=1)} RuntimeError: torch.cat(): expected a non-empty list of Tensors 原因为self.frontend.text_normalize(tts_text, split=True)返回为空	2024-07-09 08:17:34 +08:00
lyuxiang.lx	62c71075ac	update dockerfile	2024-07-08 16:40:46 +08:00
lyuxiang.lx	50c7b06ea9	compatible when ttsfrd is not avaliable	2024-07-07 13:13:42 +08:00
lyuxiang.lx	71238461f0	remove academic and change to iic/CosyVoice_ttsfrd	2024-07-07 12:19:34 +08:00
lyuxiang.lx	834053940d	update modelscope model	2024-07-06 01:53:58 +08:00
lyuxiang.lx	0379f38dd9	add cpuruntime in provider	2024-07-05 21:41:50 +08:00
lyuxiang.lx	3910efd6d3	add submodule	2024-07-04 21:40:58 +08:00
lyuxiang.lx	076829ab84	add cosyvoice code	2024-07-04 21:15:12 +08:00