Hidden conditional random fields for phone classification

TLDR

Hidden state sequences are critical for modeling the non‑stationarity of speech signals, and the HCRF framework is easily extensible to recognition as a state‑ and label‑sequence modeling technique. The paper introduces a novel application of hidden conditional random fields (HCRFs) for modeling speech. HCRFs are trained via simple stochastic gradient descent optimization, enabling efficient learning of hidden state sequences. On the TIMIT phone‑classification task, HCRFs outperform comparable ML and CML/MMI‑trained HMMs, achieving the best single‑classifier results known and handling complex features without altering the training procedure.

Abstract

In this paper, we show the novel application of hidden conditional random fields (HCRFs) – conditional random fields with hidden state sequences – for modeling speech. Hidden state sequences are critical for modeling the non-stationarity of speech signals. We show that HCRFs can easily be trained using the simple direct optimization technique of stochastic gradient descent. We present the results on the TIMIT phone classification task and show that HCRFs outperforms comparable ML and CML/MMI trained HMMs. In fact, HCRF results on this task are the best single classifier results known to us. We note that the HCRF framework is easily extensible to recognition since it is a state and label sequence modeling technique. We also note that HCRFs have the ability to handle complex features without any change in training procedure.