tipi/kyutai
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
89 Commits 38 Branches 0 Tags 1.8 MiB
af2283de3f
Commit Graph

89 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Laurent Mazare
af2283de3f
Use a streaming input in the rust example. (#102) 
* Use a streaming input in the rust example.

* Formatting.

* Another formatting tweak.
 2025-07-31 17:41:57 +02:00
Laurent Mazare
7dc926d50c
Allow for using local voices in the pytorch examples. (#100) 2025-07-31 12:48:05 +02:00
Laurent Mazare
ab8e8c59b7
Bump the version numbers. (#91) 2025-07-19 15:57:53 +02:00
laurent
5f17114618 More faq. 2025-07-18 08:43:10 +02:00
laurent
405a82ba3f FAQ tweaks. 2025-07-18 08:37:51 +02:00
Laurent Mazare
3b584b100c
Sketch a FAQ and add some issue templates. (#88) 2025-07-18 08:31:31 +02:00
Laurent Mazare
a98eb94ade
Add a streaming example for the mlx tts. (#85) 
* Add a streaming example for the mlx tts.

* Fix the CI.

* Formatting fix.

* Yet another CI fix.
 2025-07-16 22:35:44 +02:00
Laurent Mazare
a2f031deb5
Fix the pytorch tts streaming example. (#84) 
* Fix the pytorch tts streaming example.

* Edit the readme too.
 2025-07-16 21:07:02 +02:00
Laurent Mazare
66a33c989f
Add a TTS streaming example. (#83) 
* Add a TTS streaming example.

* Get the streaming example to work.
 2025-07-16 20:54:13 +02:00
laurent
89a2ced839 Bump the moshi version. 2025-07-16 16:02:28 +02:00
Laurent
baf0c75bba VAD support in the mlx-stt example that uses the microphone. 2025-07-08 16:08:32 +02:00
Laurent Mazare
952319de90
Add a MLX STT example that uses VAD. (#70) 
* Add a MLX STT example that uses VAD.

* VAD support.

* More MLX VAD example.

* Use the latest moshi-mlx.
 2025-07-08 16:04:50 +02:00
laurent
6d3bb6b1f1 Avoid the config override for the extra-heads. 2025-07-08 15:39:17 +02:00
Laurent Mazare
12dbe36b0b
Add some VAD to the pytorch speech-to-text example. (#68) 2025-07-08 11:30:34 +02:00
Václav Volhejn
cafac63222
Run pre-commit correctly in CI (#66) 
* fix and break

* Remove intentional error
 2025-07-08 10:11:52 +02:00
Václav Volhejn
7336d7a3da
Fix instructions on how to install the Rust server (#65) 
* Fix instructions on Rust server installation

* Plug Unmute
 2025-07-07 18:00:23 +02:00
Laurent Mazare
70500c620e
Add a device argument to the tts pytorch script. (#62) 2025-07-07 08:36:47 +02:00
Chenghao Mou
f8e97aa4f3
fix minor issues with readme commands (#55) 2025-07-07 08:18:05 +02:00
laurent
91a4d120cb Use moshi 0.2.8. 2025-07-07 08:12:16 +02:00
Laurent
bfc200f6ee Use bfloat16 rather than half by default. 2025-07-05 23:02:58 +02:00
laurent
f9739881e6 Typo. 2025-07-03 19:02:42 +02:00
Alexandre Défossez
99599fa408
Update README.md 2025-07-03 16:15:01 +02:00
Pierre-Hugues HUSSON
3a4165a84f
Fix stt_from_file_pytorch (#39) 
1. argparse declares in_file, but code reads file
2. text_tokens.numpy().tolist() is a list of list of list of int
instead of the supported list of list of int.
this is a debugging print just drop it

Co-authored-by: Pierre-Hugues Husson <phhusson@freebox.fr>
 2025-07-03 15:26:34 +02:00
Alexandre Défossez
e9bac066ea
Update README.md 2025-07-03 15:09:41 +02:00
Alexandre Défossez
eae5e17975
Some updates to the colab and script (#38) 
* changing streaming to be robust to repeated generation

* some changes

* plop

* plop

* plop

* plop
 2025-07-03 15:06:37 +02:00
Václav Volhejn
c1d248abba
Fix text tokenizer path (#36) 2025-07-03 14:27:06 +02:00
Václav Volhejn
c6f262346f
Don't install moshi from Git (#37) 
* Don't install moshi from Git

* Remove commented-out invalid message send in websocket_client
 2025-07-03 13:37:38 +02:00
laurent
3573ee90af Oops. 2025-07-03 13:08:00 +02:00
laurent
25574aa104 Fixes for the notebook. 2025-07-03 13:05:00 +02:00
laurent
1cd9529f65 Json fix. 2025-07-03 12:57:22 +02:00
laurent
0ee2354176 Chunk decoding in the pth notebook. 2025-07-03 12:56:00 +02:00
laurent
dc8bffabe0 Remove the dataset bit. 2025-07-03 12:48:04 +02:00
Laurent Mazare
5f8e924176
Streaming output for the pytorch tts example. (#33) 
* Streaming output for the pytorch tts example.

* Run the pre-commit hooks.
 2025-07-03 11:05:06 +02:00
Laurent Mazare
d3bed09f9a
Pin the moshi_mlx version. (#35) 2025-07-03 09:53:53 +02:00
Václav Volhejn
ef52b8ef0f
Add Rust server usage example (#32) 
* Run Ruff on tts_mlx.py

* Add tts_rust_server.py example

* Remove unused HF repo arguments and reset audio output data in TTS server script
 2025-07-03 09:47:50 +02:00
Laurent Mazare
d92e4c2695
Use the on_frame callback in the mlx tts example. (#34) 2025-07-03 09:29:04 +02:00
laurent
6c1e9f12cf Mention the MLX quantization. 2025-07-03 07:52:27 +02:00
laurent
236df522b8 Add some links. 2025-07-03 07:47:16 +02:00
laurent
20cf8d7365 Collapsible sections. 2025-07-03 07:43:56 +02:00
Laurent Mazare
ae575a04c6
Handle stdin in the mlx tts example. (#31) 2025-07-02 18:59:04 +02:00
Václav Volhejn
433dca3751
Shorter names for STT scripts (#30) 2025-07-02 18:21:42 +02:00
Václav Volhejn
07ac744609
Add PyTorch notebook and documentation (#29) 
* Add example for PyTorch implementation

* Document PyTorch and MLX examples

* Reorganize for TTS

* Remove waitlist signup CTA
 2025-07-02 17:51:27 +02:00
Václav Volhejn
96ff217437
Add Pytorch inference for TTS (#28) 
* Add tts_pytorch.py

* Add attempt at interactive playback
 2025-07-02 17:02:05 +02:00
Laurent Mazare
7294fbcc3a
Add the MLX TTS example. (#27) 
* Add the MLX TTS example.

* Audio playback.

* Allow for playing the audio in a streaming way.
 2025-07-02 16:52:26 +02:00
Eugene Kharitonov
c4ef93770a
STT example w/ prompting (#26) 
* STT example w/ prompting

* Text-audio prompt example into README.md + cutting prompt transcript.

* A line in README

* formatting in README

---------

Co-authored-by: Eugene <eugene@kyutai.org>
 2025-07-02 11:23:11 +02:00
Václav Volhejn
395eaeae95
Fix usage examples and a few small things (#24) 2025-07-02 08:58:45 +02:00
Jean-Jerome Levy
4985940aad
fix: update hf-hub dependency to 0.4.3 to resolve URL parsing error (#21) 
- Fixes RelativeUrlWithoutBase error when downloading models from fresh cache
- Resolves issue with kyutai/stt-1b-en_fr-candle model download
- Ensures proper URL handling in Hugging Face Hub interactions
 2025-06-27 13:18:27 +02:00
Václav Volhejn
0112245ef7
Refactor Rust server examples (#19) 
* Rename examples and add pre-commit

* Fix references to scripts, add implementations overview

* Link to colab notebook via github

* Simplify

* Add auth note

* Allow visualizing VAD

* Remove unused variable

* Add audio samples

* Address review comments
 2025-06-26 16:51:43 +02:00
Laurent Mazare
96eef33c4c
Add the pre-commit check to the CI. (#18) 
* Add the pre-commit check to the CI.

* CI tweaks.

* Add the pre-commit.
 2025-06-26 14:41:57 +02:00
Václav Volhejn
7b5a01dfba
Rename examples and add pre-commit (#16) 
* Rename examples and add pre-commit

* Fix references to scripts, add implementations overview

* Link to colab notebook via github
 2025-06-26 09:26:11 +02:00