Syllable-based rhythm and voice conversion of dysarthric speech, trained without labels, improves LF-MMI ASR word error rates on Torgo, especially for severe speakers, but not for fine-tuned Whisper.
As a result, Automatic Speech Recognition (ASR) systems trained on typ- ical speech often struggle to process dysarthric speech accu- rately [2]
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
eess.AS 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Unsupervised Rhythm and Voice Conversion to Improve ASR on Dysarthric Speech
Syllable-based rhythm and voice conversion of dysarthric speech, trained without labels, improves LF-MMI ASR word error rates on Torgo, especially for severe speakers, but not for fine-tuned Whisper.