An initial investigation of long-term adaptation for meeting transcription

Meeting transcription is a very useful and challenging task. The majority of research to date has focused on individual meeting, or only a small group of meetings. In many practical deployments, multiple related meetings will take place over a long period of time. This paper describes an initial investigation of how this long-term data can be used to improve meeting transcription. A corpus of technical meetings, using a single microphone array, was collected over a two year period, yielding a total of 179 hours of meeting data. Baseline systems using deep neural network acoustic models, in both Tandem and Hybrid configurations, and neural network-based language models are described. The impact of supervised and unsupervised adaptation of the acoustic models is then evaluated, as well as the impact of improved language models.