Hidden Markov Models (HMMs) have been the state-of-the-art techniques for acoustic modeling despite their unrealistic independence assumptions and the very limited representational capacity of their hidden states. There are many proposals in the research community for deeper models that are capable of modeling the many types of variability present in the speech generation process. Deep Belief Networks (DBNs) have recently proved to be very effective for a variety of machine learning problems and this paper applies DBNs to acoustic modeling. On the standard TIMIT corpus, DBNs consistently outperform other techniques and the best DBN achieves a phone error rate (PER) of 23.0 % on the TIMIT core test set. 1

In this paper, we present state-of-the art concept tagging results on a new corpus for Polish SLU. For this language, it is the first large-scale corpus (~200 different concepts) which has been semantically annotated and will be made publicly available. Conditional Random Fields have proven to lead to best results for string-to-string translation problems. Using this approach, we achieve a concept error rate of 22.6 % on an evaluation corpus. To additionally extract attribute values, a combination of a statistical and a rule-based approach is used leading to a CER of

Abstract—Gaussian mixture models are currently the dominant technique for modeling the emission distribution of hidden Markov models for speech recognition. We show that better phone recognition on the TIMIT dataset can be achieved by replacing Gaussian mixture models by deep neural networks that contain many layers of features and a very large number of parameters. These networks are first pre-trained as a multilayer generative model of a window of spectral feature vectors without making use of any discriminative information. Once the generative pre-training has designed the features, we perform discriminative fine-tuning using backpropagation to adjust the features slightly to make them better at predicting a probability distribution over the states of monophone hidden Markov models. Index Terms—Acoustic Modeling, deep belief networks, neural networks, phone recognition

Grapheme-to-Phoneme conversion (G2P) is usually used within every state-of-the-art ASR system to generalize beyond a fixed set of words. Although the performance is typically already quite good (< 10 % phoneme error rate) and pronunciations of important words are checked by a linguist, further improvements are still desirable, especially for end user customization. In this work, we present and compare five methods/tools to tackle the G2P task. Although most of the methods have already been published and/or are available as open source software, the reported experiments are done on large state-of-the-art tasks and the used software is from the actual publications. Besides an experimental comparison on text data for a range of languages (i.e. measuring the G2P accuracy only), our focus in this paper is measuring the effect of improved G2P modeling on LVCSR performance for a challenging ASR task. Additionally, the effect of using n-Best pronunciation variants instead of single best is investigated briefly. Index Terms: grapheme-to-phoneme conversion, G2P, ASR 1.