So the question remains as to how to get that last 10-15% of transcription accuracy. It sounds as if existing methodology has run out of steam and that a parad

So the question remains as to how to get that last 10-15% of transcription accuracy. It sounds as if existing methodology has run out of steam and that a paradigm shift is required to get to human-like speech recognition.