Research

Speech Communication System

  • Speech compression: for fast and efficient communication, reduce the information rate (bits/second) while maintaining the quality of the original speech waveform

 

Human Machine Interface

  • Pre-processing: Improve the quality of degraded speech for core speech signal processing such as coding and recognition
  • Recognition and classification: Recognize/classify speech-related target information (e.g. speaker, text, emotion, language, etc.) by capturing characteristics related to targets.
  • Text-to-speech: Generate synthesized speech by accepting text input.