Peer-reviewed research
Automatic Speech Recognition of Disordered Speech: Personalized models outperforming human listeners on short phrases
Jordan R. Green et al., Interspeech 2021 (Google Research publication page), 2021
Read the originalRead on October 16, 2026
- What it is
- A Google Project Euphonia study with MGH Institute of Health Professions: 432 people with self-reported disordered speech each recorded at least 300 short phrases, and personalized speech recognition models trained on their own voices were compared with general models and expert human listeners.
- How much weight it carries
- Peer-reviewed at Interspeech, the main speech technology conference. Short phrases, not long dictation, and participants were volunteers able to record hundreds of phrases, so results may not hold for everyone with a speech disorder.
Facts taken from this source
- In Google's Project Euphonia study of 432 people with disordered speech, personalized speech recognition models reached a median word error rate of 4.6 percent, against 31 percent for general speaker-independent models.
“The accuracy of personalized models for recognizing disordered speech was high (WER: 4.6%), and significantly better than speaker-independent models (WER: 31%).” (Abstract, Results)
Pages that rest on this source
- Speech to Text for Accessibility, RSI and DisabilitiesTopic
- How Speech Recognition Works, From Sound to TextGuide
- How Accurate Is Speech Recognition? The Real NumbersGuide
- How do you improve Mac dictation accuracy?Question
- Is speech recognition less accurate for some people?Question
- What is speech recognition used for?Question