One of the major problems in automatic speech recognition technologies is the sensitivity of recognizers to any interfering sounds. Since natural environments often include other sound sources, the performance of the existing technologies is severely limited. Our research team has been doing pioneering work in the recognition of sounds in mixtures, including speech, music, and environmental sounds.

 

Noise-robust automatic speech recognition

The group has studied novel recognition approaches based on sparse non-negative spectrogram representations of the noisy speech signals, which have produced state-of-the-art recognition performance in very noisy cases. In this area, the team collaborates with Radboud University Nijmegen, Carnegie Mellon University, and Aalto University speech recognition groups.

We have edited a book about noise-robust ASR, which includes contributions from world-leading researchers (Virtanen2012).