Lee M. Miller

This paper fine-tunes an ASR model on human-transcribed synthesized speech to better align its word error rate with human perception, achieving higher correlation than zero-shot ASR.

The paper uses neural entrainment measures from EEG data to predict speech-in-noise perception performance in a cohort of healthy listeners.

This study evaluates a chirped-speech stimulus (Cheech) to assess auditory evoked potentials during speech-in-noise recognition, showing it preserves intelligibility while enabling robust neural recordings.

Brain2voice 2.0 improves voice synthesis from neural signals, achieving 5.24% word error rate, 8x better than prior methods, enabling real-time intelligible speech for paralyzed individuals.