Commit graph

30 commits

Author SHA1 Message Date
Niko
65ec4d34ec
Clean up module directory layout 2025-09-29 12:33:54 -07:00
Niko
0c509ec9c5
Fix inaccurate doc comment for C++ equivalent 2025-09-02 14:18:28 -07:00
Niko
67c343082b
Remove unused error handling 2025-08-13 13:19:04 -07:00
Niko
623e5a40cb
Remove inline attributes 2025-08-13 13:17:15 -07:00
Niko
9162a2c502
Fix Clippy lint 2025-08-13 01:30:22 -07:00
Niko
814b2b836f
Make WhisperState::new unsafe 2025-08-13 01:27:28 -07:00
Niko
15e70ffd07
Refactor the entire token/segment usage
This was spurred by noticing a trivial case of UB in the original code:
all one needed was an out-of-bounds index on any of several methods with tokens or segment indexes on the state to cause UB.

I took this opportunity to consolidate methods into Rust structs that verify their index before use.
2025-08-02 18:45:42 -07:00
Niko
55e54212f1
Update whisper.cpp to v1.7.6 2025-07-30 17:04:48 -07:00
Niko
9a96b0e1c3
Convert full_get_segment_* and similar to use an internal helper instead of duplicating code 2025-02-18 17:30:53 -07:00
Niko
099faf4e2e
Convert full_get_token_* and similar to use an internal helper instead of duplicating code 2025-02-18 17:30:53 -07:00
arizhih
ec609f18ff Add support for new whisper.cpp project structure
Remove opencl feature, removed in whisper.cpp
Add openmp feature and enable by default as in whisper.cpp
Update bindings
2024-08-08 15:55:47 +02:00
arizhih
bbdc8a07ef Fix lang_detect function 2024-07-10 10:24:25 +02:00
jiahua
e4f8910b1b fix: keep &mut self 2024-04-30 09:09:52 +08:00
jiahua
f30d0ea0a1 fix: rustfmt 2024-04-26 16:23:32 +08:00
jiahua
6e04f76e41 rename WhisperInnerContext 2024-04-25 16:00:36 +08:00
jiahua
74e83185bf refactor(state): remove lifetime binding from whisper context 2024-04-25 15:30:56 +08:00
thewh1teagle
f3c86de64f
format with cargo fmt 2024-04-15 00:39:09 +03:00
thewh1teagle
ee932caf3d
feat: add get segment text lossy 2024-04-09 18:08:44 +03:00
Niko
a9e060571a
Merge remote-tracking branch 'origin/master' 2024-04-06 11:38:35 -06:00
Niko
f4ea0d97e4
Add WhisperState::full_get_token_text_lossy to ignore invalid UTF-8, instead replacing it. 2024-04-06 11:03:40 -06:00
Wenqing Zong
a026af1281 Pass fmt check 2024-03-20 16:36:08 +00:00
Wenqing Zong
384ddd77a5 Add safe bindings for speaker diarization 2024-03-20 16:30:13 +00:00
Niko
2d583d1e7f
Fix clippy lints 2024-03-18 18:57:41 -06:00
Niko
5230735222
Fix possible segmentation fault 2024-01-03 20:26:08 -07:00
travolin
bdfbeb6c60 Add access to segment text as bytes 2023-08-17 17:09:26 -07:00
jbrough
fcb477cc5e calculate mel.n_len as mel spectrogram stride len 2023-07-22 06:53:12 +01:00
Jonathan Soo
3811a77dce Change get_logits to return a single slice 2023-05-08 09:31:54 -04:00
Yuniru Yuni
ef4b9f0630
migrate state method into state object 2023-04-30 09:09:29 +09:00
Yuniru Yuni
6169229e60
refactor: delete map for State and expose struct with lifetime 2023-04-25 22:49:17 +09:00
Zero
13d44e5881
add changes from whisper.cpp update 2023-04-17 17:57:00 -06:00