Fix ElevenLabs language code extraction for multilingual ASR #107
Conversation
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
Previously, language_code was hardcoded to "eng" for all ElevenLabs transcriptions, causing poor WER on multilingual benchmarks (e.g., French: 26.34% WER, Portuguese: 35.98% WER). This fix: - Extracts language code from dataset name (e.g., "fleurs_fr" → "fr") - Dynamically sets language_code parameter based on dataset - Defaults to "en" for English-only datasets (ami, librispeech, etc.) Test results: - French: 26.34% → 3.99% WER (85% improvement) - Portuguese: 35.98% → 4.55% WER (87% improvement) 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.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.
Fixes hardcoded
language_code="eng"in ElevenLabs transcription that was causing poor performance on multilingual ASR benchmarks.Changes
extract_language_code()helper to parse language from dataset names (e.g.,fleurs_fr→fr)transcribe_with_retry()to accept and use dataset parameterResults (on small sample size)
Before fix (hardcoded "eng"):
After fix (dynamic language code):