The Best Possible Speech Recognizer on Your Own Data

Many state-of-the-art results in different machine learning areas are presented on day-to-day basis. By adjusting these systems to perform perfectly on a specific subset of all general data, huge improvements may be achieved in their resulting accuracy. Usage of domain adaptation in automatic speech recognition can bring us to production level models capable of transcribing difficult and noisy customer conversations way more accurately than the general models trained on all kinds of language and speech data. In this work I present 17% word error rate improvement in our speech recognition task over the general domain speech recognizer from Google. The improvement was achieved by both very precise annotation and preparation of domain data and by combining state-of-the-art techniques and algorithms. The described system was successfully integrated into a production environment of the Parrot transcription company, where I am a member of the initial team, which drastically increased performance of the human transcribers.

automatic speech recognition, domain data, kaldi, dataset, speech data cleaning