In some speech recognition tasks, such as man-machine dialogue systems, the spoken sentences include several recurrent phrases. A bigram language model does not adequately represent these phrases because it underestimates their probability. A better approach consists of modeling phrases as if they were individual dictionary elements. They we inserted as additional entries into the word lexicon, on which bigrams are finally computed. This paper discusses two procedures for automatically determining frequent phrases (within the framework of a probabilistic language model) in an unlabeled training set of written sentences. One procedure is optimal since it minimises the set perplexity. The other, based on information theoretic criteria, insures that the resulting model has a high statistical robustness. The two procedures are tested on a 762-word spontaneous speech recognition task. They give similar results and provide a moderate improvement over standard bigrams.

This paper describes a Continuous Speech Understanding System that allows information services to be accessed through the telephone line. It accepts queries within a restricted semantic domain, expressed in free but syntactically correct natural language, with a lexicon of the order of 800 words. In the implementation here described, a user can access an electronic mailbox or a train information service through a PABX telephone line. The architecture of the system is based on two main modules that represent and use different knowledge sources. A speaker independent recognition module generates, for each utterance, a lattice of word hypotheses which is the interface to an understanding module that performs the syntactic and semantic analysis. The recognition module is based on Hidden Markov Models of subword units, and performs the acoustic decoding process according to a beam search strategy. The understanding module finds the most likely sequence of words and represents its meaning in a format which facilitates the access to a database. It makes use of a modified caseframe analysis guided by the word hypotheses scores. Experiments were performed with 600 sentences from 10 speakers on the E-Mail application task. Using 15 Gaussian mixtures per state, a word accuracy of 75.7 was obtained with a test vocabulary of 787 words and no linguistic constraints. Linguistic processing of the corresponding lattices achieved a sentence understanding rate of 82%.

The representation of language models through stochastic finite state networks offers several attractive features for speech recognition and understanding, including the ease of integration with the algorithms used for acoustic-phonetic decoding. However, the application to real-world problems is difficult because the network sizes grow very large, and their training requires grammatical inference methods. An approach is described that keeps the network size small by avoiding detailed modeling of linguistic constructs not essential for understanding the meaning of sentences. An automatic method for training such networks starting from a corpus of semantically annotated sentences is described. The training procedure learns both the network structure and the rules for generating the semantic representation of sentences. Testing of a 787-word task achieved 92% correct sentence understanding with written input and 72% with continuous speech, speaker-independent, telephone-bandwidth spoken input.>

The authors report recent improvements to an HMM (hidden, Markov model)-based, continuous speech recognition system. These advances, which include the incorporation of interword context-dependent units and position-dependent units and an improved feature analysis, lead to a recognition system which gives a 95% word accuracy and 75% sentence accuracy for speaker independent recognition of the 1000-word, DARPA resource management task using the standard word pair grammar (with a perplexity of about 60). With the improved acoustic modeling of subword units, the overall error rate reduction was over 42% compared with the performance results reported in the baseline system. The best results obtained so far using the word pair grammar gave 95.2% average word accuracy for the three DARPA evaluation sets.>

The authors provide a detailed description of all aspects of the implementation of a large-vocabulary speaker-independent, continuous speech recognizer used as a tool for the development of recognition algorithms based on hidden Markov models (HMMs) and Viterbi decoding. The complexity of HMM recognizers is greatly increased by the introduction of detailed context-dependent units for representing interword coarticulation. A vectorized representation of the data structures involved in the decoding process, along with compilation of the connection information among temporally consecutive words and an efficient implementation of the beam search pruning, has led to a speedup of the algorithm of about one order of magnitude. A guided search can be used during a tuning phase for obtaining a speedup of more than three times. An average recognition time of about 25 s per sentence, although far from real time, allows one to perform a series of training experiments and to tune the recognition system parameters in order to obtain high word accuracy on complex recognition tasks such as the DARPA resource management task.>

Between-word context-dependent phones have been proposed to provide a more precise phonetic representation of word junctures. This technique makes it possible to accurately model soft pronunciation changes (changes in which a phone undergoes a comparatively small alteration). However, hard pronunciation changes (changes in which a phone is completely deleted or replaced by a different phone) are much less frequent and hence cannot be modeled adequately due to the lack of training material. To overcome this problem a set of phonological rules is used to redefine word junctures, specifying how to replace or delete the boundary phones according to the neighboring phones. No new speech units are required, thus avoiding most of the training issues. Results, which are evaluated on the 991-word speaker-independent DARPA task, show that phonological rules are effective in providing corrective capability at low computational cost.>

System performance is presented for a continuous speech understanding system. The system accepts natural-language queries, with a lexicon of 1000 words. Recognition and understanding work in cascade, a speaker-dependent hidden Markov model (HMM)-based word hypothesizer, generates a lattice of word candidates that is given to a grammar-based understanding stage that extracts the correct word sequence together with its meaning. The authors focus on recent improvements added at the understanding level, consisting of a new statistical modeling of correct word adjacency and function-word losses. Such modeling permits the main problems of interword coarticulation to be dealt with and produces substantial advances in terms of recognition accuracy at a low computational cost. System performance with a language model of perplexity 25 results in 87% correct sentence understanding.>

A continuous speech recognition and understanding system is presented that accepts queries about a restricted geographical domain, expressed in free but syntactically correct natural language, with a lexicon of the order of one thousand words. A lattice of word candidates hypothesized by the speaker dependent recognition level is the interface to an understanding module that performs the syntactic and semantic analysis. The recognition subsystem generates word hypotheses by exploiting hidden Markov models of sub-word units. Bottom-up constraints are also introduced to restrict the set of candidate words. The understanding module determines the most likely sequence of words and represents its meaning in a parse-tree suitable to access a database. It makes use of a modified caseframe analysis driven by the word hypotheses likelihood scores. The results of a set of experiments performed in 150 sentences collected from one speaker are given.>

The theoretical aspects of the encoding process are investigated, resulting in a precise definition of linear codes together with theorems that clarify how they can be obtained. A particular subset of linear codes, called superlinear codes, for which the performance analysis is highly simplified is identified. The most relevant performance measures for the analysis of this class of codes are discussed. The minimum Euclidean distance and the event and bit error probabilities are found analytically using the uniform error property (when applicable) or variations on it. This yields accurate upper and lower bounds to the error rate at the price of reasonable computational complexity. The theory is then applied to the search for 'good' codes and to their performance evaluation. The cases of 16- and 32-PSK codes, which are good candidates for use in digital satellite transmission, are considered. Several new results in terms of error event and bit error probabilities are presented, showing considerable gains in terms of SNR with respect to the uncoded case.>