3817f74d46
feat: nginx-based data parallel for optimal ASR throughput
...
When --dp N is specified (N > 1), the launcher now starts N independent
vLLM processes behind an nginx reverse proxy instead of using vLLM's
built-in DP coordinator. This avoids the single-process HTTP bottleneck
when handling large base64 audio payloads, achieving near-linear scaling
(7.2x with 8 GPUs at 4096 concurrent requests).
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com >
2026-03-27 07:43:32 +00:00
9634518ca4
Add data parallel (DP) support to vLLM server launcher
...
- Add --dp/--data-parallel-size flag for running independent model replicas
across multiple GPUs with automatic load balancing behind a single port
- Add --tp/--tensor-parallel-size flag (previously hardcoded to 1)
- Update docs/vibevoice-vllm-asr.md with multi-GPU deployment guide
covering DP, TP, and hybrid (DP × TP) configurations
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com >
2026-03-24 11:53:31 +00:00
165e17e5ed
fix: vllm-version-stable
2026-02-25 07:30:43 +00:00
bb54f78d0e
feat: add hotwords support for vLLM ASR
2026-02-04 10:33:20 +00:00
e43c1e2cdb
streaming use transformers==4.51.3
2026-02-03 00:30:52 -08:00
e16491d65e
Merge pull request #228 from Damon-Salvetore/vllm-1
...
[Fix] Resolve occasional infinite loops during vLLM inference
2026-02-03 10:38:40 +08:00
e26f1c263f
1
2026-02-02 13:50:27 +00:00
b2aee8015c
Delete docs/VibeVoice-ASR-Report.pdf
2026-01-28 19:33:37 +08:00
3140709188
update README
2026-01-27 21:06:31 +08:00
c435ae05d5
update README
...
Added a section on LoRA fine-tuning to the ASR documentation.
2026-01-27 21:01:40 +08:00
0e1a0d39fd
update README
2026-01-27 20:59:25 +08:00
142a00112e
update ASR README: multilingual
2026-01-27 20:58:10 +08:00
4648c50ea0
update ASR Technical Report link to Arxiv
2026-01-27 12:58:06 +08:00
a69e77c036
1. unify env for TTS and ASR; 2. avoid transformers 5.0.0 temporarily
2026-01-26 03:29:02 -08:00
1eb04f53a2
Replace install_deps.sh with start_server.py one-click deployment
2026-01-26 07:34:54 +00:00
e67b15f47d
update
2026-01-25 21:41:42 -08:00
d9068541cf
1
2026-01-25 16:11:02 +00:00
3c50e50d18
Merge pull request #203 from Damon-Salvetore/vibevoice-vllm
...
Add vLLM plugin support for high-performance ASR serving
2026-01-24 16:17:10 +08:00
7d12252de3
Language support
2026-01-24 05:11:34 +00:00
a3e99daedd
Language support
2026-01-24 05:10:47 +00:00
4df5b0582f
Add vLLM plugin support for high-performance ASR serving
2026-01-23 17:32:24 +00:00
c0c2af984e
update README for finetuning-asr
2026-01-22 06:20:11 -08:00
5022277022
update README
2026-01-22 00:51:00 -08:00
32a7040ce0
restructure README
2026-01-22 00:37:22 -08:00
ce90a49960
fix env bug
2026-01-21 22:03:52 -08:00
a3750c229b
Revise VibeVoice-ASR documentation for clarity
...
Updated the description and key features of VibeVoice-ASR to clarify its capabilities and improve accuracy in transcription.
2026-01-22 02:59:10 +08:00
c4352fee63
fx
2026-01-21 10:36:27 -08:00
616a167275
add ASR playground link
2026-01-21 10:26:17 -08:00
f7c6d2dec9
update asr eval results
2026-01-21 09:50:24 -08:00
c9c778cc58
fx
2026-01-21 08:25:53 -08:00
56cb11e7b2
Add VibeVoice-ASR
2026-01-21 22:18:33 +08:00
4adbe76674
more experimental voices
2025-12-16 04:21:09 -08:00
d295d1e1d0
Update vibevoice-realtime-0.5b.md
2025-12-09 12:28:32 +08:00
7ea24a4fb9
update
2025-12-04 22:33:57 -08:00
fc83be5d92
add VibeVoice-Realtime
2025-12-04 05:38:30 -08:00
Jianwei Yu