Researchers have developed a benchmark to test if current audio-language models can effectively use additional clinical context to improve automatic speech recognition for dysarthric speech. Initial findings indicate that these models do not significantly benefit from diagnosis labels or detailed clinical descriptions, with some prompts even degrading performance. However, fine-tuning with clinical context shows promise, achieving a substantial reduction in word error rate for specific subgroups like those with Down syndrome. AI
影响 Highlights limitations in current ASR models for atypical speech and offers a path toward more inclusive technologies.
排序理由 Academic paper presenting a new benchmark and fine-tuning method for ASR models.
AI 生成摘要 · Google Gemini · 来自 4 个来源。 我们如何撰写摘要 →
