328 Commits

Author	SHA1	Message	Date
RVC-Boss	9da7e17efe	Add files via upload	2025-04-01 18:44:35 +08:00
RVC-Boss	41090e5a7c	Update g2pw url	2025-04-01 17:15:52 +08:00
RVC-Boss	605b380114	修复模型加载异步逻辑 ... 修复模型加载异步逻辑	2025-04-01 16:50:54 +08:00
RVC-Boss	7abae557fb	删除加载v3sovits模型缺少enc_q告警 ... 删除加载v3sovits模型缺少enc_q告警	2025-04-01 16:31:15 +08:00
RVC-Boss	6a60e5edb1	v3解锁并行推理;修复模型加载异步逻辑 ... v3解锁并行推理;修复模型加载异步逻辑	2025-04-01 16:29:52 +08:00
RVC-Boss	28bdff356f	fix https://github.com/RVC-Boss/GPT-SoVITS/issues/2250 ... fix https://github.com/RVC-Boss/GPT-SoVITS/issues/2250	2025-04-01 10:34:02 +08:00
ChasonJiang	03b662a769	为sovits_v3 适配并行推理 (#2241) ... * 为sovits_v3 适配并行推理 * 清理无用代码	2025-03-31 11:56:05 +08:00
RVC-Boss	ee4a466f79	Update patched_mha_with_cache.py	2025-03-26 17:39:19 +08:00
RVC-Boss	13573a1b06	fix torch.load	2025-03-26 15:22:01 +08:00
lishq	fef65d40fe	fix: prevent concurrent access to BERT model with thread lock (#2165) ... Added thread lock to protect get_phones_and_bert method from potential race conditions during concurrent access. This addresses issue #1844 where multiple threads accessing the BERT model simultaneously could cause data inconsistency or crashes. Co-authored-by: RVC-Boss <129054828+RVC-Boss@users.noreply.github.com>	2025-03-26 15:03:36 +08:00
zzz	b0e465eb72	feat: 添加导出 v3 的 script (#2208) ... * feat: 添加导出 v3 的 script * Fix: 由于 export_torch_script_v3 的改动，v2 现在需要传入 top_k	2025-03-26 14:50:55 +08:00
KamioRinn	265586990c	更新G2PWModel下载链接 (#2219) ... * update G2PWModel download url * update G2PWModel download url	2025-03-26 14:35:52 +08:00
ChasonJiang	7394dc7b0c	为api_v2和inference_webui_fast适配V3版本 (#2188) ... * modified: GPT_SoVITS/TTS_infer_pack/TTS.py modified: GPT_SoVITS/TTS_infer_pack/TextPreprocessor.py modified: GPT_SoVITS/inference_webui_fast.py * 适配V3版本 * api_v2.py和inference_webui_fast.py的v3适配 * 修改了个远古bug,增加了更友好的提示信息 * 优化webui * 修改为正确的path * 修复v3 lora模型的载入问题 * 修复读取tts_infer.yaml文件时遇到的编码不匹配的问题	2025-03-26 14:34:51 +08:00
ChasonJiang	165882d64f	修复多余的注释导致的bug (#2158)	2025-03-05 18:22:01 +08:00
RVC-Boss	271db6a4de	fix torch.inference_mode()RuntimeError:Inplace update to inference tensor outside InferenceMode is not allowed. ... fix torch.inference_mode()RuntimeError:Inplace update to inference tensor outside InferenceMode is not allowed.	2025-03-05 18:07:47 +08:00
ChasonJiang	053a356ffe	修复gpt的padding mask的问题 (#2153) ... * 修复gpt的padding mask的问题 * rollback tts_config	2025-03-05 17:14:43 +08:00
KamioRinn	fe2f04bdb8	API for V3 (#2154)	2025-03-05 17:13:46 +08:00
ChasonJiang	6dd2f72090	更改gpt并行推理时的mask策略为padding left (#2144) ... * 更改gpt并行推理时的mask策略为padding left，使batch_infer更接近于naive_infer 减少冗余操作并使用torch_sdpa，以提升推理速度 * rollback tts_infer.yaml	2025-03-04 16:45:37 +08:00
KamioRinn	959a2ddbeb	Matching fast_langdetect update (#2140)	2025-03-04 14:10:58 +08:00
RVC-Boss	df33574a26	修复超分后音量超过1写错了的bug ... 修复超分后音量超过1写错了的bug	2025-03-01 19:17:03 +08:00
RVC-Boss	a32a2b8934	v3sovits模型推理支持webui传语速参数调整合成语速 ... v3sovits模型推理支持webui传语速参数调整合成语速	2025-02-28 16:50:35 +08:00
RVC-Boss	c38b169019	v3sovits模型推理支持webui传语速参数调整合成语速 ... v3sovits模型推理支持webui传语速参数调整合成语速	2025-02-28 16:50:12 +08:00
RVC-Boss	a69be1eae7	Merge pull request #2122 from KamioRinn/Fix-Short-CJK-LangSegmenter ... Fix Short CJK LangSegmenter	2025-02-28 11:23:05 +08:00
KamioRinn	3356fc9e09	Fix get_phones_and_bert	2025-02-28 02:25:58 +08:00
KamioRinn	531a38f119	Fix Short CJK LangSegmenter	2025-02-28 02:04:25 +08:00
starylan	e35ade8f60	拆分i18n	2025-02-28 01:41:04 +08:00
RVC-Boss	2cd843dcbc	Update inference_webui.py	2025-02-27 22:46:29 +08:00
RVC-Boss	94ffcbe616	Update inference_webui.py	2025-02-27 22:43:18 +08:00
RVC-Boss	a0ff6fa7a2	Update inference_webui.py	2025-02-27 22:28:02 +08:00
RVC-Boss	a68e3c4354	Update TTS.py	2025-02-27 22:14:51 +08:00
RVC-Boss	250b1c73cb	fix 24k to 48k inference	2025-02-27 20:49:17 +08:00
RVC-Boss	060a0d91dc	fix 24k to 48k inference ... fix 24k to 48k inference	2025-02-27 19:05:54 +08:00
RVC-Boss	00bdc01113	支持24k音频超分48k采样率 ... 支持24k音频超分48k采样率	2025-02-27 16:08:26 +08:00
RVC-Boss	00cbadcaad	屏蔽lora strict=False的print假告警，其实加载逻辑没问题。 ... 屏蔽lora strict=False的print假告警，其实加载逻辑没问题。	2025-02-27 14:38:55 +08:00
KamioRinn	237526a7e6	Fix invalid path (#2114) ... * Fix korean invalid path * Fix japanese invalid path * Fix langsegmenter invalid path	2025-02-26 23:37:19 +08:00
KamioRinn	8158a97909	Fix invalid path (#2112) ... * Fix langsegmenter invalid path * Fix langsegmenter invalid path * Fix korean invalid path	2025-02-26 18:41:52 +08:00
RVC-Boss	fbb9f21e53	Add files via upload	2025-02-23 20:26:18 +08:00
RVC-Boss	aa07216bba	修复resume epoch数识别错，每次resume都要都训一轮的问题 ... 修复resume epoch数识别错，每次resume都要都训一轮的问题	2025-02-23 15:09:22 +08:00
RVC-Boss	e6a32e15b0	support sovits v3 lora training, 8G GPU memory is enough ... support sovits v3 lora training, 8G GPU memory is enough	2025-02-23 00:38:47 +08:00
RVC-Boss	e937b625e4	support sovits v3 lora training, 8G GPU memory is enough ... support sovits v3 lora training, 8G GPU memory is enough	2025-02-23 00:37:14 +08:00
KamioRinn	a2bb1dab91	Patch-LangSegmenter (#2073)	2025-02-18 15:23:06 +08:00
KamioRinn	c17dd642c7	Add en_normalization and fix LangSegmenter (#2062)	2025-02-17 18:41:30 +08:00
wzy3650	087fd24579	fix inference issue (#2061) ... Co-authored-by: wangzeyuan <wangzeyuan@agora.io>	2025-02-17 10:38:04 +08:00
RVC-Boss	1867780d56	https://github.com/RVC-Boss/GPT-SoVITS/issues/2048 ... https://github.com/RVC-Boss/GPT-SoVITS/issues/2048	2025-02-14 12:01:27 +08:00
KamioRinn	72d839e40a	切换新的语言分割工具 (#2047) ... * Add New LangSegmenter * New LangSegmenter	2025-02-14 11:18:52 +08:00
RVC-Boss	16941a7c14	修复根据模型版本识别支持的语种滞后的问题 ... 修复根据模型版本识别支持的语种滞后的问题	2025-02-13 23:17:27 +08:00
Abyss-Seeker	87a3b908ee	Update inference_webui.py (#2043) ... 修复在Colab使用自行训练的v2模型时可能遇到的UnpicklingError: Weights only load failed.	2025-02-13 23:12:43 +08:00
RVC-Boss	d8fc921771	fix hard code	2025-02-13 11:29:31 +08:00
RVC-Boss	1b25e1097a	修复首次打开不加载sovits模型 ... 修复首次打开不加载sovits模型	2025-02-13 01:11:22 +08:00
KakaruHayate	c2b3298bed	update Gradient Checkpointing to reduce VRAM usage (#2040) ... * update Gradient Checkpointing to reduce VRAM usage * fix inference	2025-02-12 23:00:34 +08:00