Commit Graph

  • 827d6ea47c Refactor TTS and scheduler components to enhance text processing and batching capabilities. Introduce PrepareCoordinator for managing text feature preparation asynchronously, and update SchedulerDebugWorker to support new finalize task management. Implement batch processing in PrepareBertBatchWorker with improved admission control and profiling metrics. Add text CPU preprocessing utilities for better text segmentation and normalization. baicai-1145 2026-03-10 06:58:53 +08:00
  • a45e171ff5 Enhance sampling functions in TTS by adding support for previous token masks in logits_to_probs. Implement batch processing for sampling with padded token sequences and contiguous sampling groups. Refactor sampling logic in T2S scheduler to utilize new functionalities, improving efficiency and flexibility in token generation. baicai-1145 2026-03-09 21:24:16 +08:00
  • 845b181360 Implement batch processing for BERT and reference semantic tasks in TTS. Introduce StageLimiter for managing concurrent processing and enhance the TTS class with new methods for handling audio and semantic extraction. Update profiling metrics for better performance tracking during inference. baicai-1145 2026-03-09 05:19:28 +08:00
  • d245eb169c Refactor T2S scheduler and inference handling to improve attention mask management and memory tracking. Update T2SRunningRequest and T2SActiveBatch classes to include optional key padding masks. Introduce new benchmarking tools for API performance and memory usage analysis, enhancing overall system efficiency. baicai-1145 2026-03-09 01:42:04 +08:00
  • dc37b0b9ef Add WebAPI documentation and implement TTS API with endpoints for text-to-speech inference, control commands, and model switching. Enhance TTS class with methods for extracting prompt semantics and reference audio specifications. Introduce a scheduler prototype for managing T2S requests. baicai-1145 2026-03-09 00:22:59 +08:00
  • 30a4557d8d Implement last inference statistics tracking in Text2SemanticDecoder and enhance TTS class with prompt semantic extraction. This includes methods for setting and retrieving inference stats, as well as improvements to audio processing and feature extraction in TTS. baicai-1145 2026-03-08 23:08:27 +08:00
  • b250e62402 Enhance G2PW model input handling by introducing polyphonic context character support and updating the data preparation method to return additional query IDs. This improves the processing of polyphonic characters in sentences. baicai-1145 2026-03-08 03:01:20 +08:00
  • 800acd45ff Enhance G2P processing by implementing batch input handling in _g2p function, improving efficiency. Update prepare_onnx_input to utilize caching for tokenization and add optional parameters for character ID mapping and phoneme masks. Refactor G2PWOnnxConverter to streamline model loading and configuration management. baicai-1145 2026-03-07 05:47:22 +08:00
  • c0fe483288
    Merge 4820d5a101a5240b74657a55d8b75357a04d8ce3 into 2d9193b0d3c0eae0c3a14d8c68a839f1bae157dc FAN JIALI 2026-03-05 03:44:28 +00:00
  • 4820d5a101 fix: apply same Windows single-GPU gloo bypass to s2_train_v3 and s2_train_v3_lora fanfan-love-meatmeat 2026-03-05 11:44:25 +08:00
  • 832e5b6160 fix: bypass gloo DDP for Windows single-GPU training fanfan-love-meatmeat 2026-03-05 10:56:21 +08:00
  • 8b195a5adb security: replace eval() with safe boolean parsing changhaowuwu 2026-02-25 22:08:01 +01:00
  • 53b17bd2d2
    Merge pull request #1 from kaning123/Dev __kaning123__ 2026-02-25 14:01:46 +08:00
  • 69f1c9c2dd
    feat: Added path check __kaning123__ 2026-02-25 13:56:47 +08:00
  • 012eb93ef8
    feat:添加了是否启用参考音频的变量 __kaning123__ 2026-02-25 10:37:33 +08:00
  • f6e8ec8a78
    feat:Added .voice loader __kaning123__ 2026-02-25 10:20:48 +08:00
  • 1c54a945cb
    feat: Added entrys to save sv_emb and refers __kaning123__ 2026-02-25 07:53:03 +08:00
  • a6a53f7231
    feat: Added entry to disable checks __kaning123__ 2026-02-24 07:48:12 +08:00
  • a06011d838
    fix:fix import errors __kaning123__ 2026-02-23 14:29:40 +08:00
  • 6ef7c0b70f
    feat: Add lib allows tensor saving __kaning123__ 2026-02-23 09:51:55 +08:00
  • 60a25737d2
    Merge 319a09fa2d54335ab4893190c860c77beed76a83 into 2d9193b0d3c0eae0c3a14d8c68a839f1bae157dc XXXXRT666 2026-02-17 04:35:59 +00:00
  • 319a09fa2d Update Flash Attn Installation XXXXRT666 2026-02-17 04:35:40 +00:00
  • d5ed91deb5 Update Docker Scipts XXXXRT666 2026-02-17 02:46:10 +00:00
  • 0a8b006607 Update Git Ignore XXXXRT666 2026-02-17 02:03:41 +00:00
  • 76fa3426ea Partial migration to UV to speed up installation, rename the tools folder to reduce conflicts, update readme XXXXRT666 2026-02-16 22:32:23 +00:00
  • 51ada0564b
    Merge a4323233c67133f33be2db180bef07dda1376e4d into 2d9193b0d3c0eae0c3a14d8c68a839f1bae157dc e1732a364fed 2026-02-16 16:40:23 -08:00
  • 5503a5891b fix: correct torch.randint upper bound to include both values Mr-Neutr0n 2026-02-11 23:45:46 +05:30
  • 2d9193b0d3
    Migrate to miniforge, add missing dependencies, update docker file, remove deprecated files (#2732) XXXXRT666 2026-02-09 07:05:25 +00:00
  • e4ae04edf8 Add Env Vars and Secrets XXXXRT666 2026-02-09 06:40:12 +00:00
  • 3cd2ed35b2 Migrate to miniforge, add missing dependencies, update docker file, remove deprecated files XXXXRT666 2026-02-08 19:09:52 +00:00
  • 9986880b3f
    fix Conda 条款未同意导致的构建失败 (#2727) Oarora 2026-02-08 23:52:04 +08:00
  • edc9ef99ad EOS weight decay edge 2026-02-01 00:41:53 +08:00
  • f6ddd56d52 Change default top_k value to 15 for consistency Spr_Aachen 2026-01-29 10:10:10 +08:00
  • 58b891b269 fix Conda 条款未同意导致的构建失败 MeltingEtimis 2026-01-28 18:13:36 +08:00
  • d939ec3587 Create webui_correction.py yeopseung 2026-01-26 21:40:57 +09:00
  • 839d966238 chore: Update outdated GitHub Actions versions Padraic Slattery 2026-01-26 12:01:40 +01:00
  • 31824ef7a5 pull yeopseung 2026-01-26 18:17:46 +09:00
  • 7f6787121b Merge branch 'kun' of https://github.com/linguikun1986/GPT-SoVITS-Kun into kun kun 2026-01-21 00:59:08 +08:00
  • 6e027ec111 新增直接打开推理页面bat命令,针对参考音频、推理参数做了持久化配置,解决每次推理都要重复操作的痛点,新增模型记忆,即每次打开推理页面,默认加载最后一次选择的模型。如果是从主页进入,则主动加载主页选择的模修复bug (#2704) ChasonJiang 2025-12-30 16:00:21 +08:00
  • 62ee3c2aa0 feat: improve WebUI usability with presets and native dialogs Ray 2026-01-16 18:12:05 +11:00
  • ac432bb24b
    Merge 0235857b895183cae7d296389db16f3c783f189a into c767f0b83b998e996a4d230d86da575a03f54a3f hsoftxl 2026-01-14 02:36:59 +00:00
  • 0235857b89
    Merge branch 'RVC-Boss:main' into main hsoftxl 2026-01-13 17:25:10 +08:00
  • c767f0b83b
    修复bug (#2704) ChasonJiang 2025-12-30 16:00:21 +08:00
  • bd43e3cd90 fallbak and bug fix ChasonJiang 2025-12-30 15:33:15 +08:00
  • 2ed4f74741 Merge branch 'main' of https://github.com/RVC-Boss/GPT-SoVITS into fallback_fix ChasonJiang 2025-12-30 15:32:27 +08:00
  • 585305968e 修复bug ChasonJiang 2025-12-30 15:30:17 +08:00
  • 9080a967d5
    修复采样错误 (#2703) ChasonJiang 2025-12-30 15:21:03 +08:00
  • a8ea20c91e 修复采样错误 ChasonJiang 2025-12-30 14:31:11 +08:00
  • 51df9f7384
    Fix model file name in README instructions (#2700) sushistack 2025-12-25 17:44:21 +09:00
  • e325ccdf94
    Fix model file name in README instructions sushistack 2025-12-23 23:43:20 +09:00
  • bfca0f6b2d
    对齐naive_infer的解码策略,防止吞句 (#2697) ChasonJiang 2025-12-19 17:37:19 +08:00
  • 98126b4668 对齐naive_infer的解码策略,防止吞句 ChasonJiang 2025-12-19 17:17:19 +08:00
  • abe984395c
    对齐gpt topk默认采样参数 (#2696) ChasonJiang 2025-12-19 16:05:36 +08:00
  • cc89c3660e
    Update requirements.txt RVC-Boss 2025-12-19 15:54:54 +08:00
  • 8e254837f0 对齐gpt topk默认采样参数 ChasonJiang 2025-12-19 15:41:11 +08:00
  • 36b3231c6f
    bug fix (#2689) ChasonJiang 2025-12-15 14:23:06 +08:00
  • 7d01c69bc4 bug fix ChasonJiang 2025-12-13 15:41:21 +08:00
  • 9ec3a60f30
    Update config.py RVC-Boss 2025-12-01 20:23:49 +08:00
  • fc533b6fb7
    Update fasterwhisper_asr.py RVC-Boss 2025-12-01 11:38:37 +08:00
  • 857799276c
    Fix Modelscope (#2679) XXXXRT666 2025-12-01 03:13:15 +00:00
  • 61f8d2e0b0 Fix Modelscope XXXXRT666 2025-12-01 03:09:59 +00:00
  • 64214d2452
    Merge a0e4bb6ec0d9d93f3e0d345e3af33b5f5cfb4974 into 92d2d337fd98673c126fd40727e067204e4523ae XXXXRT666 2025-11-30 13:37:24 +00:00
  • a0e4bb6ec0 1 XXXXRT666 2025-11-30 13:34:34 +00:00
  • 92d2d337fd
    Fix training error caused by float type of default_batch_size parameter (#2662) Spr_Aachen 2025-11-28 22:53:43 +08:00
  • 6fb441f65e
    更友好的流模式选项 (#2678) ChasonJiang 2025-11-28 22:13:48 +08:00
  • c85c54eca9
    Add ModelScope Snapshot Download For ASR (#2627) XXXXRT666 2025-11-28 14:10:49 +00:00
  • 45e01197a3 更友好的流模式选项 ChasonJiang 2025-11-28 22:05:14 +08:00
  • cb00840c4e
    Add files via upload RVC-Boss 2025-11-28 22:02:03 +08:00
  • 60a4a214af
    vq distributed training support (#2577) wzy3650 2025-11-28 21:57:13 +08:00
  • 6375bbe316
    尝试 stream infer (#2469) zzz 2025-11-28 21:36:57 +08:00
  • e00ca92140
    Fix ASMD (#2636) KamioRinn 2025-11-28 21:22:43 +08:00
  • 92ab59c553
    更细粒度的流式推理模式 (#2671) ChasonJiang 2025-11-28 21:12:41 +08:00
  • 365760f756 修复api_v2的ogg格式传输问题 ChasonJiang 2025-11-26 15:40:34 +08:00
  • 6bce575d69 支持固定chunk长度的流式推理,优化sola算法 ChasonJiang 2025-11-26 14:41:42 +08:00
  • 9b147cd24a 更正拼写错误 ChasonJiang 2025-11-24 23:07:38 +08:00
  • fcdd15460d modified: GPT_SoVITS/TTS_infer_pack/TTS.py ChasonJiang 2025-11-24 20:52:56 +08:00
  • 08d6ed0d8c modified: GPT_SoVITS/AR/models/t2s_model.py modified: GPT_SoVITS/TTS_infer_pack/TTS.py modified: GPT_SoVITS/module/models.py modified: api_v2.py ChasonJiang 2025-11-24 20:47:32 +08:00
  • af7b95bc9d modified: .gitignore modified: GPT_SoVITS/AR/models/t2s_model.py modified: GPT_SoVITS/TTS_infer_pack/TTS.py modified: GPT_SoVITS/module/models.py ChasonJiang 2025-11-24 18:52:35 +08:00
  • d08214dd22 modified: GPT_SoVITS/TTS_infer_pack/TTS.py ChasonJiang 2025-07-01 22:27:03 +08:00
  • 0825ae80e1 modified: GPT_SoVITS/AR/models/t2s_model.py modified: GPT_SoVITS/TTS_infer_pack/TTS.py modified: GPT_SoVITS/module/models.py ChasonJiang 2025-06-24 20:43:46 +08:00
  • 9ff381b519 清理无用代码 ChasonJiang 2025-05-05 20:16:59 +08:00
  • 8e161c46fa 更好的流式推理模式 ChasonJiang 2025-05-05 20:05:29 +08:00
  • a041ee4521
    Update requirements.txt scientificworld 2025-11-23 19:34:50 +08:00
  • 8093f1156a
    Update requirements.txt scientificworld 2025-11-16 01:32:41 +08:00
  • d3b8f7e09e feat: Migrate from CUDA to XPU for Intel GPU support google-labs-jules[bot] 2025-11-10 13:09:27 +00:00
  • 35fac290ba Improve G2PWPinyin model handling - Add 'bert_pretrained_dir' environment variable as the highest priority for model_source - Replace redundant g2pw initialization in chinese2.py with import from download.py - Remove unused 'parent_directory' variable Spr_Aachen 2025-11-10 15:00:00 +08:00
  • c5d57d7afa Fix training error caused by float type of default_batch_size parameter Spr_Aachen 2025-11-10 07:02:53 +08:00
  • 1d9ab7ef60
    Merge 7338302ee4381be81acf6cf5ce7e3449a5974472 into 11aa78bd9bda8b53047cfcae03abf7ca94d27391 MiaoMiao Li 2025-11-03 07:23:30 +00:00
  • 7338302ee4 修复batch下的报错 lcc 2025-11-03 15:23:23 +08:00
  • 651ce13ef6 modified: install.sh Axiaobo 2025-10-29 03:41:01 +00:00
  • 94f157c8cf
    修复有时候会出现长段无意义音频的bug MiaoMiao Li 2025-10-28 16:12:36 +08:00
  • 8ca6bf3f11
    修复有时候会出现长段无意义音频的bug MiaoMiao Li 2025-10-28 16:08:06 +08:00
  • 7526941052 install scripts YuechuanLi 2025-10-26 13:31:37 -07:00
  • 2b93f2eb8e work with uv YuechuanLi 2025-10-26 12:51:28 -07:00
  • 2b64032cda fix MCCL error on qy1 KakaruHayate 2025-10-21 18:01:24 +08:00
  • 3a92c046f9
    revert "rfft fallback to cpu" Kakaru 2025-10-21 17:11:54 +08:00
  • 917f73c38c
    may fix RuntimeError: MUSA error: an illegal memory access was encountered Kakaru 2025-10-20 20:44:34 +08:00
  • 70a9243285
    Merge pull request #3 from plae-tljg/musa_MUSA1009 Kakaru 2025-10-20 20:43:46 +08:00
  • 42abb66a32 solved subtle problem on import and syntax plae-tljg 2025-10-20 20:28:49 +08:00
  • 06994f2a13 may fix RuntimeError: MUSA error: an illegal memory access was encountered KakaruHayate 2025-10-19 16:49:13 +08:00