Calculate duration of input files in order to decide which engine to use #136
Conversation
This file contains 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
| return parseFloat(stdout); | ||
| } catch (error) { | ||
| logger.error(`Error during ffprobe file duration detection`, error); | ||
| throw error; |
There was a problem hiding this comment.
will we return a 500 if we can't find file duration here? might be better to default to whisperx
There was a problem hiding this comment.
good shout, implemented in 5d7709a
packages/common/src/types.ts
Outdated
| @@ -69,6 +69,8 @@ export const TranscriptionJob = z.object({ | |||
| engine: z.nativeEnum(TranscriptionEngine), | |||
| // we can get rid of this when we switch to using a zip | |||
| translationOutputBucketUrls: z.optional(OutputBucketUrls), | |||
| // this is optional because giant currently doesn't know file duration | |||
There was a problem hiding this comment.
do ffmpeg or the whisper models output file duration? Or would it be worth having the worker calculate duration with ffprobe? That would unify the way we're finding duration and make it so Giant isn't a special case.
There was a problem hiding this comment.
Good point. I was more just shoving this in in case useful for logging - the only place it was used was in setting a 'duration' field in the transcription failure message, which isn't really necessary. So I've removed this field in 244a6d2
Deploy build 801 of
|
244a6d2 to
1833874
Compare
zekehuntergreen
approved these changes
Feb 24, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
What does this change?
Currently we have two available 'engines' for transcription:
It seems a shame to wait to spin up whisperx for a file that is only a few minutes long. This PR gets the API to make a decision on which engine to use based on the duration of the file - any files under 10 minutes long where the user hasn't requested diarization will be sent to whisper.cpp, everything else goes to whisperX
To get ffprobe running on lambda I had to create a lambda layer - this is basically just a zip file that gets added to the disk of the lambda vm.
How to test
Tested on CODE