Speaker adaptation of neural network acoustic models using i-vectors

• Published in: 2013 IEEE Workshop on Automatic Speech Recognition and Understanding pp. 55 - 59
• Main Authors: Saon, George, Soltau, Hagen, Nahamoo, David, Picheny, Michael
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-12-2013
• Subjects: Acoustics; Feature extraction; Hidden Markov models; Neural networks; Training; Training data; Vectors
• DOI: 10.1109/ASRU.2013.6707705

We propose to adapt deep neural network (DNN) acoustic models to a target speaker by supplying speaker identity vectors (i-vectors) as input features to the network in parallel with the regular acoustic features for ASR. For both training and test, the i-vector for a given speaker is concatenated to every frame belonging to that speaker and changes across different speakers. Experimental results on a Switchboard 300 hours corpus show that DNNs trained on speaker independent features and i-vectors achieve a 10% relative improvement in word error rate (WER) over networks trained on speaker independent features only. These networks are comparable in performance to DNNs trained on speaker-adapted features (with VTLN and FMLLR) with the advantage that only one decoding pass is needed. Furthermore, networks trained on speaker-adapted features and i-vectors achieve a 5-6% relative improvement in WER after hessian-free sequence training over networks trained on speaker-adapted features only.