Show live segments for the CTranslate2 whisper engine (looked frozen)#12239
Merged
niksedk merged 1 commit intoJul 7, 2026
Merged
Conversation
whisper-ctranslate2 prints no segment lines unless --verbose is on; its only other progress output is a carriage-return tqdm bar that never completes a line, so Subtitle Edit's newline-based console log stayed completely empty for the whole run. With large-v3 decoding sequentially on the CPU (macOS has no GPU path in CTranslate2) that silence lasts a very long time and the engine looks frozen or stuck in a loop, when it is in fact transcribing. Default --verbose True into the CTranslate2 parameter set, unless the user already set their own --verbose preference in advanced settings. Segments now stream into the console as they are decoded, exactly like the other engines. Verified on macOS against the shipped whisper-ctranslate2 bundle and the Systran large-v3 model: without the flag the console stays empty until the end; with it, "[MM:SS.mmm --> MM:SS.mmm] text" lines appear live.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
CTranslate2 engine: console stays empty for the whole run, looks frozen
Problem
Running the Whisper CTranslate2 engine, the console log shows the "Calling speech-to-text" line and then nothing at all until the run finishes. whisper-ctranslate2 prints no segment lines unless --verbose is on; its only other progress output is a carriage-return tqdm bar, which never completes a line, so the newline-based console log never gets anything to show.
On macOS this is especially misleading: CTranslate2 has no Apple GPU backend, so a large-v3 decode runs sequentially on the CPU and takes a long time, all of it in total silence. It looks frozen or stuck in a loop while it is actually transcribing (that is how it was reported).
Fix
Default
--verbose Trueinto the CTranslate2 parameter set, unless the user already set their own--verbosepreference in the advanced parameters. Segments then stream into the console as they are decoded, matching the behavior of the other engines.Verified
On macOS against the shipped whisper-ctranslate2 0.5.7 bundle and the Systran large-v3 model, with the exact command line Subtitle Edit builds:
[MM:SS.mmm --> MM:SS.mmm] textlines appear live per segment (Arabic text included)Files
src/ui/Features/Video/SpeechToText/SpeechToTextViewModel.cs