Qwen3-TTS-streaming

Nighthawk/Qwen3-TTS-streaming

Fork 0

mirror of https://github.com/Nighthawk42/Qwen3-TTS-streaming.git synced 2026-08-30 08:42:26 +00:00

611f1022d4 Added additional endpoints, updated Gradio demo. main Nighthawk 2026-02-18 01:48:10 -05:00
b0588a0c93 Added initial API code and documentation. Nighthawk 2026-02-18 01:38:09 -05:00
b8db6ff1ce update README.md: list experimental branch features; GPU-resident repetition penalty and batch compaction Kedara Studios 2026-02-12 18:43:22 +01:00
a757bf2bf0 update README.md -> move to experimental branch Kedara Studios 2026-02-12 02:36:47 +01:00
81275c6895 Update README.md Kedara Studios 2026-02-10 11:21:01 +01:00
c7b9bc6b76 fix: disable repetition penalty by default for streaming and add windowed penalty Kedara Studios 2026-02-10 10:43:06 +01:00
0cc89f619c fix: apply upstream ConvTranspose padding and audio_lengths fixes Kedara Studios 2026-02-09 21:22:57 +01:00
233a141057 revert: restore codec levels 1-15 embedding summing in finetuning Kedara Studios 2026-02-09 20:05:23 +01:00
32c00eb088 fix: remove double label-shifting in finetuning that caused speech speed-up Kedara Studios 2026-02-09 19:46:43 +01:00
a48510a58d refactor: batch GPU decode calls in batch_stream_generate_voice_clone Kedara Studios 2026-02-09 01:57:30 +01:00
dc7a8da2b1 fix: guard chunk trimming against zero blend_samples to prevent full truncation Kedara Studios 2026-02-08 19:16:34 +01:00
74fd5823c5 fix: guard fade-out against zero blend_samples to prevent full-array fade Kedara Studios 2026-02-08 18:43:41 +01:00
7e616bbf9f docs: reorder README sections and update repetition_penalty description Kedara Studios 2026-02-08 17:45:00 +01:00
fa20ec2dc3 docs: add batch streaming section to README with usage example Kedara Studios 2026-02-08 17:32:22 +01:00
c6d0a2c416 Merge pull request #8 from rekuenkdr/feature/batch-streaming rekuenkdr 2026-02-08 17:30:24 +01:00
117ec6192a feat: add batch streaming generation for parallel multi-item TTS synthesis Kedara Studios 2026-02-08 17:23:39 +01:00
1275736834 Merge pull request #7 from rekuenkdr/fix/repetition-penalty-and-finetune-fixes rekuenkdr 2026-02-07 10:42:41 +01:00
f88566ac7e Merge branch 'main' into fix/repetition-penalty-and-finetune-fixes rekuenkdr 2026-02-07 10:42:24 +01:00
a3f95c5e65 docs: add repetition_penalty param to README with streaming context Kedara Studios 2026-02-07 10:36:34 +01:00
cecfe01b67 Merge pull request #6 from rekuenkdr/fix/repetition-penalty-and-finetune-fixes rekuenkdr 2026-02-06 16:33:30 +01:00
837960761b chore: relax transformers and accelerate version pins to >= minimum Kedara Studios 2026-02-06 16:31:24 +01:00
1a067a0cc6 Merge pull request #5 from rekuenkdr/fix/repetition-penalty-and-finetune-fixes rekuenkdr 2026-02-06 16:26:41 +01:00
f83f18439d fix: add repetition penalty to streaming and sync upstream finetuning Kedara Studios 2026-02-06 16:25:05 +01:00
97da215d4d Merge pull request #4 from rekuenkdr/fix/audio-crossfade-fixes rekuenkdr 2026-02-04 19:38:18 +01:00
3640831a47 fix: handle multiple EOS tokens for generation termination Kedara Studios 2026-02-04 19:34:58 +01:00
d9724dd494 Merge pull request #3 from rekuenkdr/fix/audio-crossfade-fixes rekuenkdr 2026-02-03 08:56:13 +01:00
909e1089be feat(audio): make overlap_samples fully configurable Kedara Studios 2026-02-03 08:48:13 +01:00
ec5c320a41 Merge pull request #2 from rekuenkdr/fix/audio-crossfade-fixes rekuenkdr 2026-02-02 23:39:15 +01:00
96137074c8 fix(audio): Hann window crossfade and boundary click prevention Kedara Studios 2026-02-02 23:38:16 +01:00
92aaeac93c Merge pull request #1 from rekuenkdr/feature/two-phase-emit rekuenkdr 2026-02-02 20:29:27 +01:00
7bf2c43a2e feat: add two-phase streaming for reduced first-chunk latency Kedara Studios 2026-02-02 20:27:15 +01:00
01b51f0cb1 README update dffdeeq 2026-01-28 02:26:50 +04:00
8c533a60e0 Merge pull request #3 from dffdeeq/fix/streaming-startup-artifacts Andrew 2026-01-27 19:49:39 +04:00
b7f12c4941 fix(streaming): add ref_code context for stable first chunks dffdeeq 2026-01-27 19:14:52 +04:00
990bfa2766 Revert upstream changes that broke inference speed dffdeeq 2026-01-27 11:18:11 +04:00
e5662324c6 minor dffdeeq 2026-01-26 23:24:18 +04:00
66f26eb640 Merge pull request #2 from dffdeeq/feature/inference-speed-up Andrew 2026-01-26 23:17:05 +04:00
ee3bf7a862 Merge branch 'main' into feature/inference-speed-up Andrew 2026-01-26 23:16:53 +04:00
b5a75200d9 reference changed dffdeeq 2026-01-26 23:14:50 +04:00
648b7100bd benchmarks update Andrew 2026-01-26 17:54:30 +04:00
45d455dd3f Merge pull request #1 from dffdeeq/feature/inference-speed-up Andrew 2026-01-26 16:57:34 +04:00
f2ae2b039c no streaming optimization dffdeeq 2026-01-26 15:52:39 +04:00
d8488d2727 ,,, dffdeeq 2026-01-26 06:16:51 +04:00
e03d44f3f4 README.md update Andrew 2026-01-25 23:45:27 +04:00
d7fd4bb1ae Update language and text for voice cloning example Andrew 2026-01-25 23:33:58 +04:00
ec708743ea Merge branch 'QwenLM:main' into main Andrew 2026-01-25 23:26:36 +04:00
1ab0dd7535 update: update readme Xiong Wang 2026-01-26 00:11:02 +08:00
3b30a4e509 update: update readme Xiong Wang 2026-01-24 21:44:35 +08:00
c25ce958ea Merge pull request #15 from vasqu/fix-fa-flags Xiong Wang 2026-01-24 16:32:32 +08:00
ab0f77849b fix loading logic vasqu 2026-01-23 10:48:11 +01:00
587ebf6389 fixup wrong flags vasqu 2026-01-23 10:18:08 +01:00
0c6a7cbb6c update: modify default non_streaming_mode Xiong Wang 2026-01-23 14:08:26 +08:00
6c07ea2a85 Merge branch 'QwenLM:main' into main Andrew 2026-01-23 09:20:45 +04:00
8a98526da0 fix and update: fix finetuning/prepare_data.py and update citation Xiong Wang 2026-01-23 13:19:19 +08:00
2331298c38 add streaming TTS generation dffdeeq 2026-01-23 08:58:38 +04:00
d5c446171a fix: hf speech_tokenizer download bug. Xiong Wang 2026-01-22 23:34:21 +08:00
d8146d5305 Initial commit Xiong Wang 2026-01-22 17:57:04 +08:00

Commit Graph Select branches Hide Pull Requests main Mono Color

Commit Graph

Select branches

Hide Pull Requests

main