A new study published on arXiv compares the performance of human listeners and three advanced automatic speech recognition (ASR) systems—Whisper-large-V3, Google Chirp 3, and Omnilingual—in recognizing Dutch dysarthric continuous speech. The research found that both humans and the off-the-shelf ASR systems struggled significantly, with word error rates (WER) exceeding 70% on average. However, fine-tuning the ASR models on dysarthric speech led to a substantial reduction in WER, with personalized models outperforming human listeners and showing promise for supporting daily communication. AI
IMPACT Personalized ASR models show potential to improve daily communication for individuals with dysarthria, though significant challenges remain.
RANK_REASON The cluster contains an academic paper detailing research findings.
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
