Far-Field Automatic Speech Recognition

The machine recognition of speech spoken at a distance from the microphones, known as far-field automatic (ASR), has received significant increase in attention science and industry, which caused or was by an equally improvement accuracy. Meanwhile, it entered consumer market with digital home assistants language interface being its most prominent application. Speech recorded is affected various acoustic distortions, consequently, quite different processing pipelines have emerged compared ASR for close-talk speech. A signal enhancement front end dereverberation, source separation, beamforming employed to clean up speech, back-end engine robustified multicondition training adaptation. We will also describe so-called end-to-end approach ASR, new promising architecture that recently been extended scenario. This tutorial article gives account algorithms used enable accurate distance, be seen that, although deep learning share technological breakthroughs, clever combination traditional can lead surprisingly effective solutions.