We describe a new approach to speech recognition, in which all Hidden Markov Model (HMM) states share the same Gaussian Mixture Model (GMM) structure with the same number of Gaussians in each state. The model is defined by vectors associated with each state with a dimension of, say, 50, together with a global mapping from this vector space to the space of parameters of the GMM. This model appears to give better results than a conventional model, and the extra structure offers many new opportunities for modeling innovations while maintaining compatibility with most standard techniques.

No abstract

Although research has previously been done on multilingual speech recognition, it has been found to be very difficult to improve over separately trained systems. The usual approach has been to use some kind of "universal phone set" that covers multiple languages. We report experiments on a different approach to multilingual speech recognition, in which the phone sets are entirely distinct but the model has parameters not tied to specific states that are shared across languages. We use a model called a "Subspace Gaussian Mixture Model" where states' distributions are Gaussian Mixture Models with a common structure, constrained to lie in a subspace of the total parameter space. The parameters that define this subspace can be shared across languages. We obtain substantial WER improvements with this approach, especially with very small amounts of inlanguage training data.

We investigate a number of Deep Neural Network (DNN) architectures for emotion identification with the IEMOCAP database. First we compare different feature extraction frontends: we compare high-dimensional MFCC input (equivalent to filterbanks), versus frequency-domain and time-domain approaches to learning filters as part of the network. We obtain the best results with the time-domain filter-learning approach. Next we investigated different ways to aggregate information over the duration of an utterance. We tried approaches with a single label per utterance with time aggregation inside the network; and approaches where the label is repeated for each frame. Having a separate label per frame seemed to work best, and the best architecture that we tried interleaves TDNN-LSTM with time-restricted self-attention, achieving a weighted accuracy of 70.6%, versus 61.8% for the best previously published system which used 257-dimensional Fourier log-energies as input.

No abstract

Preparation of a lexicon for speech recognition systems can be a significant effort in languages where the written form is not exactly phonetic. On the other hand, in languages where the written form is quite phonetic, some common words are often mispronounced. In this paper, we use a combination of lexicon learning techniques to explore whether a lexicon can be learned when only a small lexicon is available for boot-strapping. We discover that for a phonetic language such as Spanish, it is possible to do that better than what is possible from generic rules or hand-crafted pronunciations. For a more complex language such as English, we find that it is still possible but with some loss of accuracy.

In this paper we present a novel approach for estimating featurespace maximum likelihood linear regression (fMLLR) transforms for full-covariance Gaussian models by directly maximizing the likelihood function by repeated line search in the direction of the gradient. We do this in a pre-transformed parameter space such that an approximation to the expected Hessian is proportional to the unit matrix. The proposed algorithm is as efficient or more efficient than standard approaches, and is more flexible because it can naturally be combined with sets of basis transforms and with full covariance and subspace precision and mean (SPAM) models.

We propose to exploit phone posteriors as an additional feature in Deep Neural Network (DNN) to recognize emotions from raw speech waveform. The proposed DNN setup uses a time domain approach of learning filters within the network. The frame-level phone posteriors are combined with the learned feature representation through the network. Appended learned time domain features and phone posteriors are used as an input to the temporal context modeling layers which interleaves TDNN-LSTM with time-restricted self-attention. We achieve 16.48% relative error rate improvement in IEMOCAP categorical problem (with a final weighted accuracy of 75.03%) using phone posteriors compared to DNN setup which uses only learned time domain features for temporal context modeling. Further, we study the effect of learning emotion categories leveraging dimensional primitives in multi-task learning DNN model.