valenti/spark-vibevoice
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
123 Commits 1 Branch 0 Tags
b9d561240ada3ee5d8fb5812bebb32f7ecfd97ae
Commit Graph
123 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
Xinyang Qiu b9d561240a Add inference steps and seed controls to Gradio demo (#40) 2025-12-23 20:57:21 -08:00
Ludwittge 1407c3315c fix: update Colab link to point to new repository (#43) 2025-12-23 15:27:19 -08:00
mrfakename d150f453d5 Pin gradio version as gradio 6 breaks it 2025-12-08 20:41:26 -08:00
mrfakename e0bef6f8b2 Fix Gradio demo 2025-12-06 14:48:43 -08:00
mrfakename 2b672dd405 Bump version for streaming 2025-12-06 14:45:52 -08:00
mrfakename 864a4a7ed2 Add VibeVoice streaming model (#35) 2025-12-04 22:00:30 -08:00
mrfakename 187cf62036 Add support for loading Safetensors LoRAs 2025-10-27 08:51:49 -07:00
mrfakename db4cad0723 Fix for windows 2025-10-22 14:32:46 -07:00
cseti007 63a21e2b45 add ---seed parameter for reproducibility (#18) 2025-10-04 19:25:35 -07:00
mrfakename 7e2b324801 Allow passing a custom tqdm class 2025-10-02 16:15:25 -07:00
mrfakename c25ae58bb2 LoRA inference (#14) 2025-09-24 15:05:02 -07:00
mrfakename 7ea18d2eac Force EOS token, pad audio (#15) 2025-09-24 15:02:41 -07:00
mrfakename 9f37b409ed Update FINETUNING.md 2025-09-23 14:06:59 -07:00
Etherll ae6cb88889 Make is_prefill configurable (#11) 2025-09-23 11:39:53 -07:00
mrfakename 74d57a72a9 Update FINETUNING.md 2025-09-23 08:59:46 -07:00
mrfakename a84b1003d6 Update README.md 2025-09-23 08:52:27 -07:00
mrfakename 60f8873718 Update FINETUNING.md 2025-09-23 08:50:07 -07:00
mrfakename f1b697c157 Update Discord support link with additional resource 
Added a link to the VibeVoice-finetuning repository for reference.
 2025-09-22 18:23:36 -07:00
aqachun f22fc38448 docs: Update model path references to VibeVoice-7B (#8) 2025-09-18 18:00:22 -07:00
mrfakename ea94d5b03d New loss function for training (#9) 2025-09-18 08:06:17 -07:00
mrfakename 00a868c095 Mark training as done 2025-09-16 20:03:48 -07:00
mrfakename ed52bd539b Add LoRA fine-tuning scripts (#5) 2025-09-16 21:03:14 -07:00
mrfakename d244401c9d Update README.md 2025-09-14 08:20:15 -07:00
mrfakename 4954a3f2d6 Community repo cleanup 2025-09-07 16:34:31 -07:00
mrfakename 9a8c5249a4 Update README.md 2025-09-06 13:49:43 -07:00
mrfakename 00db115965 Update links and model weights in README 2025-09-05 09:34:27 -07:00
mrfakename 8ddecb87b1 Revise README with latest VibeVoice updates 
Updated README with recent VibeVoice developments and links.
 2025-09-05 09:30:49 -07:00
mrfakename bd04f975b6 Swap link out 2025-09-04 13:44:41 -07:00
mrfakename 9bbbe964cc Update README.md 2025-09-04 09:00:22 -07:00
mrfakename 7b1fa0a4f9 Update README.md 2025-09-04 09:00:01 -07:00
YaoyaoChang 6065c5224e update Colab: add download audio 2025-09-02 02:08:19 -07:00
pengzhiliang 21c35b4701 fix bug: from_pretrained of VibeVoiceProcessor 2025-09-01 11:39:59 +00:00
YaoyaoChang 5d09c31021 update readme 2025-09-01 18:02:28 +08:00
gregory-fanousandYaoyaoChang 6dd49cb720 feat: add Apple MPS device support (dtype + attention handling) to demos (#67) 
* Enhance model loading with device support and error handling

Updated device handling for model loading and added support for MPS. Improved error handling and fallback mechanisms for attention implementations.

* Improve device handling and model loading logic

Updated device argument handling to support MPS and added validation for MPS availability. Enhanced model loading logic based on the selected device type.

* fallback only when flash_attention_2 and add some comments back

---------

Co-authored-by: YaoyaoChang <cyy574006791@qq.com>
 2025-09-01 17:16:01 +08:00
YaoyaoChang 3074f898ca support CPU 2025-09-01 16:32:50 +08:00
jonathanagustin 086cd6aa6c refactor(colab): remove cd and use absolute paths for idempotent runs 
- avoid stateful working-directory changes in notebook cells
- use explicit /content/... paths for clone/install/inference
- improves readability and re-run reliability
 2025-09-01 15:47:52 +08:00
jonathanagustin 12493567c4 perf(colab): simplify setup & speed up downloads (uv, hf-xet) 
- use `uv pip -e` for faster, deterministic installs
- enable hf-xet (HF_XET_HIGH_PERFORMANCE=1) for faster transfers
- download model with `hf download` to /content/models/VibeVoice-1.5B (quiet, idempotent)
- simplify flow: absolute paths, minimal cells, remove redundant steps
- add binary T4 GPU check + clear runtime switch instructions
- add “Open in Colab” badge; keep Risks & Limitations visible
 2025-09-01 15:47:52 +08:00
pengzhiliang 01a3005bd9 update README 2025-09-01 06:27:15 +00:00
pengzhiliang 29433e0f27 update README 2025-09-01 06:23:52 +00:00
YaoyaoChang 174e53fc04 gradio support sdpa 2025-08-30 08:22:32 -07:00
Rafael Milewski 999020e7c4 add dark mode version 2025-08-29 17:25:43 +08:00
Z c0194a7a00 Specifically reference newer llvmlite and numba dependencies 
llvmlite is trying to build an old version that creates a dependency chain issue when installing with uv:
```
  help: `llvmlite` (v0.36.0) was included because `vibevoice` (v0.0.1) depends on `librosa` (v0.11.0) which
        depends on `numba` (v0.53.1) which depends on `llvmlite`
```

pinning these dependencies on newer versions forces it to resolve and shouldn't have any downsides.
 2025-08-29 16:19:01 +08:00
Sangjee Dondrub 4d5d9779c8 Remove redundant librosa package in the depdencies. 2025-08-29 15:28:31 +08:00
Before Crash ccdb19f004 update requires-python = '>=3.9' (#48) 2025-08-29 15:25:33 +08:00
JianweiYu f3e35d0c2c update 2025-08-28 21:58:31 -07:00
JianweiYu 6210e6ee68 update 2025-08-28 21:50:31 -07:00
JianweiYu 1a75187097 update 2025-08-28 18:40:51 -07:00
JianweiYu cd78c5b6e6 update 2025-08-28 18:29:18 -07:00
Li Dong b815902059 Merge pull request #28 from DevMello/patch-1 
fix: remove typos in README.md
 2025-08-29 00:01:37 +08:00
Li Dong 174fbabbf2 Merge pull request #45 from RichardAtCT/patch-1 
docs: fix spelling errors in README.md
 2025-08-29 00:01:11 +08:00