Hasty Briefsbeta

Bilingual

Silent speech with ultrasound

6 hours ago
  • #silent-speech
  • #speech-decoding
  • #ultrasound-tongue-imaging
  • Researchers developed a model to predict speech from ultrasound tongue images during silent speech, achieving a 15.6% word error rate on open-vocabulary speech.
  • The system uses a 50-hour dataset collected from audible speech, as tongue movements are similar between silent and vocalized speech, enabling quality-check with audio transcription.
  • It employs ResNet-18 2+1d for video encoding and Whisper decoder for text prediction, with training focused on aligning video embeddings to Whisper's audio embeddings.
  • The model generalizes to new speakers with American accents but struggles with non-American accents, and performance improves with more data without signs of plateauing.
  • Hardware challenges include miniaturizing the ultrasound probe and replacing gel with hydrogel, aiming for a wearable or adhesive patch for practical daily use.