8th European Conference on Speech Communication and Technology

Geneva, Switzerland
September 1-4, 2003


Adapting Acoustic Models to New Domains and Conditions Using Untranscribed Data

Asela Gunawardana, Alex Acero

Microsoft Research, USA

This paper investigates the unsupervised adaptation of an acoustic model to a domain with mismatched acoustic conditions. We use techniques borrowed from the unsupervised training literature to adapt an acoustic model trained on the Wall Street Journal corpus to the Aurora-2 domain, which is composed of read digit strings over a simulated noisy telephone channel. We show that it is possible to use untranscribed in-domain data to get significant performance improvements, even when it is severely mismatched to the acoustic model training data.

Full Paper

Bibliographic reference.  Gunawardana, Asela / Acero, Alex (2003): "Adapting acoustic models to new domains and conditions using untranscribed data", In EUROSPEECH-2003, 1633-1636.