Speaker-independent Continuous Speech Recognition System Research Articles

Investigation Amazigh speech recognition using CMU tools

The aim of this paper is to describe the development of a speaker-independent continuous automatic Amazigh speech recognition system. The designed system is based on the Carnegie Mellon University Sphinx tools. In the training and testing phase an in house Amazigh_Alphadigits corpus was used. This corpus was collected in the framework of this work and consists of speech and their transcription of 60 Berber Moroccan speakers (30 males and 30 females) native of Tarifit Berber. The system obtained best performance of 92.89 % when trained using 16 Gaussian Mixture models.

A Syllable Lattice Approach to Speaker Verification

This paper proposes a syllable-lattice-based speaker verification algorithm for Mandarin Chinese input. For each speech utterance, a syllable lattice is generated with a speaker-independent large-vocabulary continuous speech recognition system in free syllable decoding. The verification decision is made based upon the likelihood ratio between a target-speaker model and a speaker-independent background model, computed on the decoded syllable lattice. The likelihood function is calculated efficiently in a forward algorithm by considering all paths in the lattice. The proposed algorithm was evaluated using a Mandarin Chinese database, where 1832 true and 26 250 impostor trials were recorded by 19 target speakers and 180 impostors. The average duration of each trial is 2 s long without silence. The target-speaker model was adapted from the speaker-independent background model using enrollment data of two minutes with silence. The proposed algorithm achieved an equal-error rate of 0.857% which is better than 1.21% of the hidden Markov model-based speaker verification algorithm without using syllable lattices. The equal-error rate was further reduced to 0.617% by incorporating the Goussian mixture model-universal background model algorithm with 2048 Gaussian kernels whose equal error rate is 0.990%.

Can continuous speech recognizers handle isolated speech?

Continuous speech is far more natural and efficient than isolated speech for communication. However, for current state-of-the-art automatic speech recognition systems, isolated speech recognition (ISR) is far more accurate than continuous speech recognition (CSR). It is common practice in the speech research community to build CSR systems using only CS data. However, slowing of the speaking rate is a natural reaction for a user faced with the high error rates of current CSR systems. Ironically, CSR systems typically have a much higher word error rate when speakers slow down since the acoustic models are usually derived exclusively from continuous speech corpora. In this paper, we summarize our efforts to improve the robustness of our speaker-independent CSR system against speaking styles, without suffering a recognition accuracy penalty. In particular the multi-style trained system described in this paper attains a 7.0% word error rate for a test set consisting of both isolated and continuous speech, in contrast to the 10.9% word error rate achieved by the same system trained only on continuous speech.

Context-dependent connectionist probability estimation in a hybrid hidden Markov model-neural net speech recognition system

In this paper we present a training method and a network architecture for estimating context-dependent observation probabilities in the framework of a hybrid hidden Markov model (HMM)/multi layer perceptron (MLP) speaker-independent continuous speech recognition system. The context-dependent modeling approach we present here computes the HMM context-dependent observation probabilities using a Bayesian factorization in terms of context-conditioned posterior phone probabilities which are computed with a set of MLPs, one for every relevant context. The proposed network architecture shares the input-to-hidden layer among the set of context dependent MLPs in order to reduce the number of independent parameters. Multiple states for phone models with different context dependence for each state are used to model the different context effects at the beginning and end of phonetic segments. A new training procedure that "smooths" networks with different degrees of context dependence is proposed to obtain a robust estimate of the context-dependent probabilities. We have used this new architecture to model generalized biphone phonetic contexts. Tests with the speaker-independent DARPA Resource Management database have shown average reductions in word error rates of 28% using a word-pair grammar, compared to our earlier context-independent HMM/MLP hybrid.

Automated voice dictation in French

Designing a Voice-Activated Typewriter in French necessitates a study both on how to design the acoustic level recognition, and on how to obtain a model of the French language. Such a project was initiated at LIMSI 15 years ago. This paper presents the different steps that have been completed since the beginning of this project. First, a study on the phoneme-to-grapheme conversion, for continuous, error-free phonemic strings, using a large vocabulary and a natural language syntax was completed in 1979. The corresponding results were then improved, with attempts to convert phoneme strings containing (simulated) errors, while the methodology was adapted to the case of stenotype-to-grapheme conversion. In the ESPRIT project 860 “Linguistic Analysis of the European Languages”, our approach for language modeling was compared with other approaches on 7 different European languages. The link between the acoustic recognition and the language model resulted in a complete system (“Hamlet”), for a limited vocabulary (2,000 words), pronounced in isolation, which was then extended to a vocabulary of 5,000 words, taking advantage of a specialized DTW chip (MuPCD), also designed at LIMSI. This study resulted in the conclusion that dictation in an isolated mode was not acceptable. A speaker-independent continuous speech recognition system is now developed for vocabularies of 5 to 20 KWords.

A speaker-independent continuous speech recognition system using continuous mixture Gaussian density HMM of phoneme-sized units

The author describes a large vocabulary, speaker-independent, continuous speech recognition system which is based on hidden Markov modeling (HMM) of phoneme-sized acoustic units using continuous mixture Gaussian densities. A bottom-up merging algorithm is developed for estimating the parameters of the mixture Gaussian densities, where the resultant number of mixture components is proportional to both the sample size and dispersion of training data. A compression procedure is developed to construct a word transcription dictionary from the acoustic-phonetic labels of sentence utterances. A modified word-pair grammar using context-sensitive grammatical parts is incorporated to constrain task difficulty. The Viterbi beam search is used for decoding. The segmental K-means algorithm is implemented as a baseline for evaluating the bottom-up merging technique. The system has been evaluated on the TIMIT database (1990) for a vocabulary size of 853. For test set perplexities of 24, 104, and 853, the decoding word accuracies are 90.9%, 86.0%, and 62.9%, respectively. For the perplexity of 104, the decoding accuracy achieved by using the merging algorithm is 4.1% higher than that using the segmental K-means (22.8% error reduction), and the decoding accuracy using the compressed dictionary is 3.0% higher than that using a standard dictionary (18.1% error reduction). >

On large-vocabulary speaker-independent continuous speech recognition

In this paper, we describe sphinx, the world's first accurate large-vocabulary speaker-independent continuous speech recognition system. We will present current results of sphinx, compare its performance against similar systems, and account for its high accuracy.

Burst point location in stop consonants using backpropagation neural networks

A three-layered neural network approach for burst point location is presented, which can be used to extract consonant segments in a speaker-independent continuous speech recognition system. By using neural networks trained with the backpropagation algorithm [Ruemlhart et al., Nature 323, 533–536 (1986)], nonlinearity is introduced into the articulatory event detection decision making. The system can detect the burst point location in French voiced stop consonants /b,d,g/. For the experiments, a neural network structure of 12–20 units in the hidden layer, 50 units in the input layer, and 1 unit in the output layer is used. The input patterns represent the time series values of the speech power transition. The network was trained in several steps initially using a smaller set of training data and then using larger sets. The results of the burst location detection were encouraging, especially for the syllables “ba,” “dou,” and “ga.” Generally, the detection rate for /b/ was two times better than that for /d/ and /g/. There was no remarkable difference between the detection rates in the training data and in the unknown data.