df44c234 |
2023-11-29 02:06:23 |
add env |
Le Wang |
04972782 |
2023-11-29 06:20:12 |
add .env |
Le Wang |
209ebc7e |
2023-11-28 09:58:38 |
add requirements for openaiapi |
Le Wang |
6c9ebd04 |
2023-11-28 09:53:13 |
add an openai-like api |
Le Wang |
65094b77 |
2023-11-27 11:12:47 |
Update info on cuBLAS and cuDNN libs in README.md (#513) |
Purfview |
9641d5f5 |
2023-11-27 02:43:35 |
Force read-mode in `av.open` (#566) |
Clayton Yochum |
e1a218fa |
2023-11-24 23:19:47 |
Bump version to 0.10.0 |
Dang Chuan Nguyen |
30844096 |
2023-11-24 23:16:12 |
Add V3 Support (#578) |
Oscaarjs |
5a0541ea |
2023-09-18 16:21:37 |
Bump version to 0.9.0 |
Guillaume Klein |
e94711bb |
2023-09-14 17:42:02 |
Add property WhisperModel.supported_languages (#476) |
Guillaume Klein |
0048844f |
2023-09-14 17:17:01 |
Expose function available_models (#475) |
Guillaume Klein |
a49097e6 |
2023-09-12 15:45:54 |
Add some missing typing annotations in transcribe.py |
Guillaume Klein |
81086f6d |
2023-09-12 14:44:37 |
Always run the encoder at the beginning of the loop (#468) |
Guillaume Klein |
f6979456 |
2023-09-12 14:44:22 |
Update tokenizers requirement to include version 0.14 (#469) |
Guillaume Klein |
727ab81f |
2023-09-12 10:02:23 |
Improve error message for invalid task and language parameters (#466) |
Guillaume Klein |
0285d46f |
2023-09-08 14:35:17 |
Add more details about the requirements in the README (#463) |
Guillaume Klein |
ad388cd3 |
2023-09-04 11:56:48 |
Bump version to 0.8.0 |
Guillaume Klein |
4a41746e |
2023-09-04 11:55:40 |
Log a warning when the model is English-only but the language is set to something else (#454) |
Guillaume Klein |
1e6eb967 |
2023-09-04 11:54:42 |
Add "large" alias for "large-v2" model (#453) |
Guillaume Klein |
f0ff1296 |
2023-09-01 17:31:30 |
Expose generation parameter no_repeat_ngram_size (#449) |
Guillaume Klein |
5871858a |
2023-09-01 15:25:13 |
Force the garbage collector to run after decoding the audio with PyAV (#448) |
Guillaume Klein |
e87fbf8a |
2023-08-31 10:19:48 |
Added audio duration after VAD to TranscriptionInfo object (#445) |
MinorJinx |
7b271da0 |
2023-08-17 12:20:24 |
docs: add wscribe to community integrations (#427) |
Hrishikesh Barman |
1562b023 |
2023-08-06 05:08:24 |
added repetition_penalty to TranscriptionOptions (#403) |
Aisu Wata |
1ce16652 |
2023-08-04 08:06:17 |
Adds DEBUG log message for prompt_reset_on_temperature (#399) |
Purfview |
857be6f6 |
2023-08-03 17:44:37 |
Rename clear_previous_text_on_temperature argument (#398) |
Purfview |
1a1eb1a0 |
2023-08-03 22:40:58 |
Add clear_previous_text_on_temperature parameter (#397) |
KH |
5c17de17 |
2023-07-24 11:10:12 |
Bump version to 0.7.1 |
Guillaume Klein |
0f55c436 |
2023-07-24 10:57:15 |
Invalidate the cached encoder output when no_speech threshold is met (#376) |
Guillaume Klein |
e786e26f |
2023-07-20 23:13:11 |
Return result with best log prob when all temperature fallbacks failed (#356) |
KH |
687db319 |
2023-07-18 23:03:01 |
Remove duplicate code (#359) |
KH |
171d90dd |
2023-07-18 15:23:47 |
Bump version to 0.7.0 |
Guillaume Klein |
0e051a5b |
2023-07-18 15:22:39 |
Prepend prefix tokens with the initial timestamp token (#358) |
Guillaume Klein |
2a37390f |
2023-07-18 15:08:53 |
Minor reformatting in code snippet |
Guillaume Klein |
3b4a6aa1 |
2023-07-05 22:16:53 |
Improve timestamp heuristics (#336) |
Hoon |
c7cb2aa8 |
2023-07-03 23:40:10 |
Add support for using whisper models from Huggingface by specifying the model id. (#334) |
zh-plus |
c0d93d08 |
2023-07-03 10:20:36 |
Avoid computing higher temperatures on no_speech segments (#225) |
Guillaume Klein |
19c294f9 |
2023-07-03 10:20:20 |
Squash long words at window and sentence boundaries (#226) |
Guillaume Klein |
fee52c92 |
2023-06-21 20:46:20 |
Allow users to input an Iterable of token ids into initial_prompt (#306) |
FlippFuzz |
efc4f61d |
2023-06-20 10:53:11 |
Do not specify the vocabulary file extension in the download pattern (#311) |
Guillaume Klein |
ad58ba26 |
2023-06-16 14:37:45 |
Fix typo (#304) |
kh |
20d4e941 |
2023-06-10 14:22:29 |
Add Open-Lyrics as a community project. (#291) |
zh-plus |
d4222da9 |
2023-06-07 11:30:53 |
Update README with community repo using FW (#284) |
Antonio Zarauz Moreno |
1bb7e33b |
2023-05-24 18:22:44 |
Reformat code snippet in README |
Guillaume Klein |
2a006215 |
2023-05-24 16:15:01 |
Bump version to 0.6.0 |
Guillaume Klein |
a150adcc |
2023-05-24 16:07:54 |
Enable onnxruntime dependency for Python 3.11 (#260) |
Guillaume Klein |
ae1e6d98 |
2023-05-24 15:56:03 |
Remove reference to the VAD function from the README |
Guillaume Klein |
cf7c0215 |
2023-05-24 15:50:37 |
Export __version__ at the module level (#258) |
Guillaume Klein |
4db549b8 |
2023-05-24 15:49:36 |
Make get_speech_timestamps backward compatible with the previous usage (#259) |
Guillaume Klein |
c99feb22 |
2023-05-24 12:55:15 |
Include requirements files in sdist (#240) |
Guillaume Klein |
723cb974 |
2023-05-24 12:55:04 |
Fix occasional IndexError on empty segments (#227) |
Guillaume Klein |
6a2da9a9 |
2023-05-11 15:07:15 |
Also catch client-side network exceptions when synchronizing models (#228) |
Guillaume Klein |
6a1d331d |
2023-05-11 15:06:46 |
Add CONTRIBUTING.md (#229) |
Guillaume Klein |
2d7c984b |
2023-05-11 14:47:22 |
Reformat function download_model for clarity |
Guillaume Klein |
8e5c747a |
2023-05-11 12:15:41 |
Reformat list of community integrations |
Guillaume Klein |
32b962be |
2023-05-09 19:20:41 |
Adds: whisper-standalone-win (#216) |
Purfview |
53d247b0 |
2023-05-09 11:20:22 |
retry model download locally if huggingface throws an http error. (#215) |
David Axelrod |
91f948b0 |
2023-05-09 13:53:47 |
transcribe: return all language probabilities if requested (#210) |
Ozan Caglayan |
5d8f3e2d |
2023-05-09 18:47:02 |
Implement VadOptions (#198) |
FlippFuzz |
d889345e |
2023-04-28 10:56:13 |
added whisper-diarize (#193) |
Mahmoud Ashraf |
5d203d27 |
2023-04-27 14:53:28 |
Update Github link to community project (#187) |
Jordi Mas |
a3dcb900 |
2023-04-26 17:38:16 |
Bump version to 0.5.1 |
Guillaume Klein |
89a4c7f1 |
2023-04-26 17:37:51 |
Update docstring to clarify download_root and output_dir |
Guillaume Klein |
6f9d68dd |
2023-04-26 17:36:24 |
Fix typing of local_files_only |
Guillaume Klein |
68df3214 |
2023-04-26 16:35:18 |
Use cache_dir instead of local_dir (#182) |
Jordi Mas |
67cce3f5 |
2023-04-25 17:00:41 |
Bump version to 0.5.0 |
Guillaume Klein |
8340e04d |
2023-04-25 15:54:31 |
Assign words to the speech chunk with the greatest coverage (#180) |
Guillaume Klein |
8cf5d5a4 |
2023-04-25 15:54:22 |
Increase the default value of speech_pad_ms to 400 ms (#179) |
Guillaume Klein |
32dc625f |
2023-04-25 15:47:38 |
Update README.md |
Guillaume Klein |
e06511f9 |
2023-04-24 16:29:17 |
Rename AudioInfo to TranscriptionInfo (#174) |
Guillaume Klein |
338a725f |
2023-04-24 16:28:47 |
fix where the tokens are reset (#175) |
Anthony |
f8931137 |
2023-04-24 09:04:42 |
Align segment structure with openai/whisper (#154) |
Amar Sood |
2b51a97e |
2023-04-24 21:02:19 |
Add transcription_options to AudioInfo (#170) |
FlippFuzz |
358d3736 |
2023-04-20 14:26:06 |
Allow specifying local_files_only to prevent checking the Internet everytime (#166) |
Jordi Mas |
3adcc12d |
2023-04-13 09:50:53 |
Clarify that the returned segments value is a generator (#144) |
Guillaume Klein |
2b53dee6 |
2023-04-08 10:02:36 |
Expose download location in WhisperModel constructor (#126) |
Ewald Enzinger |
06d24056 |
2023-04-06 20:13:09 |
Configure ignore for more files. (#122) |
Bekir Bakar |
e9a082dc |
2023-04-06 11:54:40 |
Keep segment timestamps aligned with words timestamps after VAD (#119) |
Guillaume Klein |
051b3350 |
2023-04-05 16:57:59 |
Add some info and debug logs (#113) |
Guillaume Klein |
746f2698 |
2023-04-04 12:16:23 |
Bump version to 0.4.1 |
Guillaume Klein |
a5d03e55 |
2023-04-04 10:51:14 |
Prevent out of range error in method split_tokens_on_unicode (#111) |
Guillaume Klein |
9fa19890 |
2023-04-04 10:25:41 |
Revert "Prevent out of range error in method split_tokens_on_unicode" |
Guillaume Klein |
36160c1e |
2023-04-04 10:17:56 |
Prevent out of range error in method split_tokens_on_unicode |
Guillaume Klein |
2f266eb8 |
2023-04-03 19:34:54 |
Fix VAD index error when a predicted timestamps is too large (#107) |
Guillaume Klein |
8c36ac1b |
2023-04-03 17:24:49 |
Bump version to 0.4.0 |
Guillaume Klein |
19698c95 |
2023-04-03 17:22:48 |
Support VAD filter (#95) |
Guillaume Klein |
b4c1c577 |
2023-04-03 22:56:35 |
Added retrieval mechanism (avg_log_prob/no_speech_prob) (#103) |
palladium123 |
f20bb258 |
2023-04-03 11:22:43 |
Support separating the left and right audio channels (#97) |
Guillaume Klein |
1a968a43 |
2023-04-01 09:26:42 |
Pass prefix only to the first window |
Guillaume Klein |
def70d84 |
2023-03-31 18:54:55 |
Update headings in the Usage section |
Guillaume Klein |
7301df7f |
2023-03-31 17:06:44 |
Update README.md (#101) |
mayeaux |
d03383f9 |
2023-03-30 15:58:27 |
Simplify reuse of the encoder output |
Guillaume Klein |
39fddba8 |
2023-03-30 12:42:29 |
Suppress some special tokens when the default set is not used |
Guillaume Klein |
eda840f8 |
2023-03-29 12:11:24 |
Always disable the progress bar specific to snapshot_download |
Guillaume Klein |
02244005 |
2023-03-28 14:36:10 |
Add large-v1 model |
Guillaume Klein |
8246479f |
2023-03-27 10:19:22 |
Ignore the invalid audio frames (#82) |
Guillaume Klein |
e2705d11 |
2023-03-26 16:29:11 |
Raise an explicit error message if the model size is invalid |
Guillaume Klein |
f8d2fb16 |
2023-03-25 10:00:59 |
Fix variable name reference (#77) |
Jordi Mas |
a10732c7 |
2023-03-24 17:59:11 |
Only download the required model files |
Guillaume Klein |
7808eddf |
2023-03-24 10:56:42 |
Bump version to 0.3.0 |
Guillaume Klein |
de7682a2 |
2023-03-24 10:55:55 |
Automatically download converted models from the Hugging Face Hub (#70) |
Guillaume Klein |
523ae218 |
2023-03-24 10:53:49 |
Run the encoder only once for each 30-second window (#73) |
Guillaume Klein |
2b7be470 |
2023-03-24 09:15:05 |
Update README.md |
Guillaume Klein |
3f02c536 |
2023-03-23 20:52:46 |
Add .gitignore file |
Guillaume Klein |
e663186a |
2023-03-23 20:33:19 |
Add some badges at the top of the README |
Guillaume Klein |
e44a8c7b |
2023-03-22 21:07:27 |
Update the README following the PyPI release |
Guillaume Klein |
33f41d84 |
2023-03-22 21:01:53 |
Add job to push a package for each new Git tag |
Guillaume Klein |
c910ec02 |
2023-03-22 20:54:07 |
Bump version to 0.2.0 |
Guillaume Klein |
e9dfe23e |
2023-03-22 20:53:51 |
Complete the package metadata |
Guillaume Klein |
66efd02b |
2023-03-22 20:50:03 |
Run some automatic tests with GitHub Actions (#68) |
Guillaume Klein |
52264f22 |
2023-03-22 13:51:12 |
Fix typing for device_index argument |
Guillaume Klein |
c27c010f |
2023-03-21 17:13:37 |
Ignore Unicode errors in input file metadata |
Guillaume Klein |
0ab8db2b |
2023-03-18 09:48:02 |
Remove debug prints |
Guillaume Klein |
a70aac18 |
2023-03-18 09:47:02 |
Remove unused import |
Guillaume Klein |
d82be59d |
2023-03-17 18:33:16 |
Fix unset attribute when using English-only models |
Guillaume Klein |
58f44479 |
2023-03-17 16:44:07 |
Update benchmark results with latest openai/whisper and faster-whisper |
Guillaume Klein |
cce6b53e |
2023-03-16 10:32:36 |
Fix incorrect attribute access |
Guillaume Klein |
2007adf0 |
2023-03-15 17:49:07 |
Fix typing of words attribute |
Guillaume Klein |
ae9898f0 |
2023-03-15 15:30:29 |
Include duration in AudioInfo structure |
Guillaume Klein |
c5f6b91b |
2023-03-15 15:27:20 |
Port utility function format_timestamp |
Guillaume Klein |
eafb2c79 |
2023-03-15 15:22:53 |
Add more typing annotations |
Guillaume Klein |
8bd013ea |
2023-03-15 15:02:28 |
Add word-level timestamps (#43) |
Guillaume Klein |
b41fd059 |
2023-03-10 11:15:58 |
Update python_requires to >=3.8 |
Guillaume Klein |
3301dd92 |
2023-03-09 12:54:41 |
Make get_input a free function |
Guillaume Klein |
c52adaca |
2023-03-09 12:53:49 |
Create a helper class Tokenizer |
Guillaume Klein |
f0a21ea9 |
2023-03-09 11:53:55 |
Use a dict to represent intermediate segments |
Guillaume Klein |
6a84df40 |
2023-03-09 10:02:25 |
Fix all_tokens handling |
Guillaume Klein |
4176da0d |
2023-03-09 09:58:58 |
Rename offset to seek to match the OpenAI implementation |
Guillaume Klein |
6b16b8a6 |
2023-03-08 10:50:46 |
Pad the audio instead of the spectrogram |
Guillaume Klein |
26469065 |
2023-03-07 10:15:36 |
Fix error in decode_audio for long audio inputs |
Guillaume Klein |
01ef12a6 |
2023-03-07 10:05:04 |
Do not ignore last segment ending with one timestamp |
Guillaume Klein |
469244a5 |
2023-03-06 16:21:48 |
Update CTranslate2 to 3.8.0 |
Guillaume Klein |
4a18adc3 |
2023-03-01 15:47:16 |
Load the tokenizer from the model directory if it exists |
Guillaume Klein |
87399262 |
2023-02-28 19:01:31 |
Accept the audio waveform as an input to transcribe() (#21) |
Guillaume Klein |
ed32002a |
2023-02-27 12:21:54 |
Add instructions to install without git clone |
Guillaume Klein |
a4f1cc8f |
2023-02-27 12:09:40 |
Add prefix parameter |
Guillaume Klein |
528aa3e7 |
2023-02-27 11:32:03 |
Make threshold parameters optional |
Guillaume Klein |
f0add58b |
2023-02-27 11:22:02 |
Add typing to constructor and transcribe method |
Guillaume Klein |
b1c69927 |
2023-02-24 15:52:23 |
Update code snippet to be consistent with the conversion example |
Guillaume Klein |
ef71be09 |
2023-02-23 11:18:58 |
Update CTranslate2 to 3.7.0 |
Guillaume Klein |
f5c0e449 |
2023-02-22 14:59:29 |
Update README.md |
Guillaume Klein |
d91365e3 |
2023-02-22 11:02:11 |
Minor code simplification |
Guillaume Klein |
4b8237da |
2023-02-22 10:28:04 |
Strip the leading space before computing the compression ratio |
Guillaume Klein |
e47e0091 |
2023-02-22 10:27:38 |
Add length_penalty parameter and correctly compute the avg log prob |
Guillaume Klein |
f5c9f15c |
2023-02-21 12:10:54 |
Check that the language code is valid |
Guillaume Klein |
a98a2eee |
2023-02-17 18:51:12 |
Use the large model in the GPU benchmark |
Guillaume Klein |
8321fcb9 |
2023-02-17 14:42:09 |
Recompute the performance numbers on GPU |
Guillaume Klein |
e2094b64 |
2023-02-17 14:37:24 |
Reduce the maximum length when the prompt is longer than 448/2 |
Guillaume Klein |
5b240319 |
2023-02-16 17:38:58 |
Update benchmark results with ctranslate2==3.6.0 |
Guillaume Klein |
123d9a57 |
2023-02-16 17:02:40 |
Support English-only models |
Guillaume Klein |
cda834c8 |
2023-02-16 17:01:19 |
Update CTranslate2 to 3.6.0 |
Guillaume Klein |
0b535499 |
2023-02-14 17:54:50 |
Add whisper.cpp in benchmark table |
Guillaume Klein |
17a6d83d |
2023-02-14 16:58:05 |
Add some performance numbers in the README |
Guillaume Klein |
cbbe6330 |
2023-02-14 09:34:05 |
Add num_workers parameter |
Guillaume Klein |
c86353d3 |
2023-02-13 21:26:25 |
Add task parameter |
Guillaume Klein |
f56dfc64 |
2023-02-13 21:22:05 |
Add without_timestamps parameter |
Guillaume Klein |
5e938cba |
2023-02-13 21:16:54 |
Bump minimum CTranslate2 requirement to 3.5.1 |
Guillaume Klein |
3dc44f7b |
2023-02-13 18:26:45 |
Raise a more explicit error message for English-only models |
Guillaume Klein |
47a62ab9 |
2023-02-13 17:43:22 |
Update README.md |
Guillaume Klein |
90f6923b |
2023-02-13 16:08:31 |
Update code snippet to output seconds as float |
Guillaume Klein |
269b3dfb |
2023-02-13 11:06:40 |
Expose the device_index argument (#5) |
Guillaume Klein |
0bcbbfa8 |
2023-02-12 12:05:30 |
Update README.md |
Guillaume Klein |
3e7b8109 |
2023-02-12 12:04:11 |
Add not about GPU requirements |
Guillaume Klein |
60e667e0 |
2023-02-12 11:44:05 |
Cleanup unused import |
Guillaume Klein |
7d1d0541 |
2023-02-12 11:42:21 |
Add the initial_prompt parameter (#2) |
Guillaume Klein |
23d2d642 |
2023-02-11 11:47:07 |
Update transcribe.py |
Guillaume Klein |
c0ec7fe8 |
2023-02-11 11:46:09 |
Update README.md |
Guillaume Klein |
5216d52d |
2023-02-11 10:21:19 |
Initial commit |
Guillaume Klein |