Show Thoughts

A system that stores multiple trained speech recognition models for different environments and retrieves them guided by visual scene recognition has improved speech recognition in reverberated and noisy environments.