sam/kdenlive-transcript-sync
Watch
1
0
Fork
You've already forked kdenlive-transcript-sync
 0
Code Issues Pull requests Projects Releases Packages 1 Wiki Activity Actions

feat(whisper): Add device, compute type, and batch size options #4

Merged
sam merged 1 commit from thread-and-cuda into main 2026-08-16 18:23:26 +02:00
Conversation 0 Commits 1 Files changed 4 +390 -110
sam commented 2026-08-16 18:20:01 +02:00
Owner
Copy link

Add configuration options for faster-whisper inference tuning: -
--whisper-device: Select inference device (auto/cpu/cuda) -
--whisper-compute-type: Set ctranslate2 compute type -
--whisper-batch-size: Enable batched GPU decoding

Implement transcribe_many() for prefetched parallel audio
decoding. Refactor model caching to include device/compute-type in
key. Add device and compute-type resolution helpers. Improve error
handling with fallback to cpu/int8. Update config documentation and
CLI help text.

Add configuration options for faster-whisper inference tuning: - `--whisper-device`: Select inference device (auto/cpu/cuda) - `--whisper-compute-type`: Set ctranslate2 compute type - `--whisper-batch-size`: Enable batched GPU decoding Implement `transcribe_many()` for prefetched parallel audio decoding. Refactor model caching to include device/compute-type in key. Add device and compute-type resolution helpers. Improve error handling with fallback to cpu/int8. Update config documentation and CLI help text.
feat(whisper): Add device, compute type, and batch size options
All checks were successful
Arch Package / build-publish (pull_request) Has been skipped
Details
Arch Package / manual-build-publish (pull_request) Has been skipped
Details
Arch Package / test-build (pull_request) Successful in 2m46s
Details
Arch Package / test-build (push) Has been skipped
Details
Arch Package / manual-build-publish (push) Has been skipped
Details
Arch Package / build-publish (push) Successful in 2m41s
Details
e2b4eaa062 
Add configuration options for faster-whisper inference tuning: -
`--whisper-device`: Select inference device (auto/cpu/cuda) -
`--whisper-compute-type`: Set ctranslate2 compute type -
`--whisper-batch-size`: Enable batched GPU decoding

Implement `transcribe_many()` for prefetched parallel audio
decoding.  Refactor model caching to include device/compute-type in
key. Add device and compute-type resolution helpers. Improve error
handling with fallback to cpu/int8. Update config documentation and
CLI help text.
sam merged commit e2b4eaa062 into main 2026-08-16 18:23:26 +02:00
sam deleted branch thread-and-cuda 2026-08-16 18:23:26 +02:00
Sign in to join this conversation.
Reviewers
No reviewers
Labels
Clear labels
No items
No labels
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
sam/kdenlive-transcript-sync!4
No description provided.