Degradation Of Recognition Performance Research Articles

Influence of sampling on target recognition and identification

Two perception experiments are conducted to quantify the relationship between imager sampling artifacts and target recognition and identification performance using that imager. The results of these experiments show that in-band aliasing (aliasing that overlaps the baseband signal) does not degrade target identification performance, but out- of-band aliasing (such as visible display raster) degrades identification performance significantly. Aliasing had less impact on the recognition task than the identification task, but both in-band and out-of-band aliasing moderately degrades recognition performance. Based on these experiments and other results reported in the literature, it appears that in-band aliasing has a strong effect on low-level discrimination tasks such as point (hot-spot) detection; out-of-band aliasing has only a minor impact on these tasks. For high-level discrimination tasks such as target identification, however, out-of-band aliasing has a significant impact on performance, whereas in-band aliasing has a minor affect. For intermediate-level discrimination tasks such as target recognition, both in-band and out-of-band aliasing have a moderate impact on performance. Based on data from the perception experiments, the modulation transferfunction (MTF) model is developed. The degraded performance due to undersampling is modeled as an effective increase in system blur or, equivalently, a contraction or squeeze in the MTF. An equation is developed that quantifies the amount of MTF or contraction to apply to the system MTF to account for the performance degradation caused by sampling.

Listening to two simultaneous speeches

Speech stream segregation is presented as a new speech enhancement for automatic speech recognition. Two issues are addressed: speech stream segregation from a mixture of sounds, and interfacing speech stream segregation with automatic speech recognition. Speech stream segregation is modeled as a process of extracting harmonic fragments, grouping these extracted harmonic fragments, and substituting non-harmonic residue for non-harmonic parts of a group. The main problem in interfacing speech stream segregation with hidden Markov model (HMM)-based speech recognition is how to improve the degradation of recognition performance due to spectral distortion of segregated sounds, which is caused mainly by transfer function of a binaural input. Our solution is to re-train the parameters of HMM with training data binauralized for four directions. Experiments with 500 mixtures of two women's utterances of an isolated word showed that the error reduction rate of the 1-best/10-best word recognition of each woman's utterance is, on average, 64% and 75%, respectively.

Maximum-likelihood stochastic-transformation adaptation of hidden Markov models

The recognition accuracy in previous large vocabulary automatic speech recognition (ASR) systems is highly related to the existing mismatch between the training and testing sets. For example, dialect differences across the training and testing speakers result in a significant degradation in recognition performance. Some popular adaptation approaches improve the recognition performance of speech recognizers based on hidden Markov models with continuous mixture densities by using linear transformations to adapt the means, and possibly the covariances of the mixture Gaussians. The linear assumption, however, is too restrictive, and in this paper we propose a novel adaptation technique that adapts the means and, optionally, the covariances of the mixture Gaussians by using multiple stochastic transformations. We perform both speaker and dialect adaptation experiments, and we show that our method significantly improves the recognition accuracy and the robustness of our system. The experiments are carried out with SRI's DECIPHER speech recognition system.

Pattern recognition method

One-dimensional normal distributions in respective dimensions of a continuous multi-dimensional normal distribution of each state of HMMs representing speech units mean and variance values are tied among similar one-dimensional distributions. As a result, the total number of normal distributions for representing the model is reduced without degrading recognition performance.

Fast and accurate acoustic modelling with semi-continuous HMMs

In this paper the design of accurate Semi-Continuous Density Hidden Markov Models (SC-HMMs) for acoustic modelling in large vocabulary continuous speech recognition is presented. Two methods are described to improve drastically the efficiency of the observation likelihood calculations for the SC-HMMs. First, reduced SC-HMMs are created, where each state does not share all the – gaussian – probability density functions ( pdfs) but only those which are important for it. It is shown how the average number of gaussians per state can be reduced to 70 for a total set of 10 000 gaussians. Second, a novel scalar selection algorithm is presented reducing to 5% the number of gaussians which have to be calculated on the total set of 10 000, without any degradation in recognition performance. Furthermore, the concept of tied state context-dependent modelling with phonetic decision trees is adapted to SC-HMMs. In fact, a node splitting criterion appropriate for SC-HMMs is introduced: it is based on a distance measure between the mixtures of gaussian pdfs as involved in SC-HMM state modelling. This contrasts with other criteria from literature which are based on simplified pdfs to manage the algorithmic complexity. On the ARPA Resource Management task, a relative reduction in word error rate of 8% was achieved with the proposed criterion, comparing with two known criteria based on simplified pdfs.

A frequency warping approach to speaker normalization

In an effort to reduce the degradation in speech recognition performance caused by variation in vocal tract shape among speakers, a frequency warping approach to speaker normalization is investigated. A set of low complexity, maximum likelihood based frequency warping procedures have been applied to speaker normalization for a telephone based connected digit recognition task. This paper presents an efficient means for estimating a linear frequency warping factor and a simple mechanism for implementing frequency warping by modifying the filterbank in mel-frequency cepstrum feature analysis. An experimental study comparing these techniques to other well-known techniques for reducing variability is described. The results have shown that frequency warping is consistently able to reduce word error rate by 20% even for very short utterances.

Analysis and compensation of speech under stress and noise for environmental robustness in speech recognition

It is well known that the introduction of acoustic background distortion and the variability resulting from environmentally induced stress causes speech recognition algorithms to fail. In this paper, several causes for recognition performance degradation are explored. It is suggested that recent studies based on a Source Generator Framework can provide a viable foundation in which to establish robust speech recognition techniques. This research encompasses three inter-related issues: (i) analysis and modeling of speech characteristics brought on by workload task stress, speaker emotion/stress or speech produced in noise (Lombard effect), (ii) adaptive signal processing methods tailored to speech enhancement and stress equalization, and (iii) formulation of new recognition algorithms which are robust in adverse environments. An overview of a statistical analysis of a Speech Under Simulated and Actual Stress (SUSAS) database is presented. This study was conducted on over 200 parameters in the domains of pitch, duration, intensity, glottal source and vocal tract spectral variations. These studies motivate the development of a speech modeling approach entitled Source Generator Framework in which to represent the dynamics of speech under stress. This framework provides an attractive means for performing feature equalization of speech under stress. In the second half of this paper, three novel approaches for signal enhancement and stress equalization are considered to address the issue of recognition under noisy stressful conditions. The first method employs (Auto:I,LSP:T) constrained iterative speech enhancement to address background noise and maximum likelihood stress equalization across formant location and bandwidth. The second method uses a feature enhancing artificial neural network which transforms the input stressed speech feature set during parameterization for keyword recognition. The final method employs morphological constrained feature enhancement to address noise and an adaptive Mel-cepstral compensation algorithm to equalize the impact of stress. Recognition performance is demonstrated for speech under a range of stress conditions, signal-to-noise ratios and background noise types.

A maximum-likelihood approach to stochastic matching for robust speech recognition

Presents a maximum-likelihood (ML) stochastic matching approach to decrease the acoustic mismatch between a test utterance and a given set of speech models so as to reduce the recognition performance degradation caused by distortions in the test utterance and/or the model set. We assume that the speech signal is modeled by a set of subword hidden Markov models (HMM) /spl Lambda//sub x/. The mismatch between the observed test utterance Y and the models /spl Lambda//sub x/ can be reduced in two ways: 1) by an inverse distortion function F/sub /spl nu//(.) that maps Y into an utterance X that matches better with the models /spl Lambda//sub x/ and 2) by a model transformation function G/sub /spl eta//(.) that maps /spl Lambda//sub x/ to the transformed model /spl Lambda//sub x/ that matches better with the utterance Y. We assume the functional form of the transformations F/sub /spl nu//(.) or G/sub /spl eta//(.) and estimate the parameters /spl nu/ or /spl eta/ in a ML manner using the expectation-maximization (EM) algorithm. The choice of the form of F/sub /spl nu//(.) or G/sub /spl eta//(.) is based on prior knowledge of the nature of the acoustic mismatch. The stochastic matching algorithm operates only on the given test utterance and the given set of speech models, and no additional training data is required for the estimation of the mismatch prior to actual testing. Experimental results are presented to study the properties of the proposed algorithm and to verify the efficacy of the approach in improving the performance of a HMM-based continuous speech recognition system in the presence of mismatch due to different transducers and transmission channels.

Signal conditioning techniques for robust speech recognition

Acoustic mismatch encountered in various training and testing conditions of hidden Markov model (HMM) based systems often causes severe degradation in speech recognition performance. For telephone based speech recognition tasks, acoustic mismatch can arise from various sources, such as variations in telephone handsets, ambient noise, and channel distortions. This paper presents three techniques for blind channel equalization, namely, cepstral mean subtraction (CMS), signal bias removal (SBR) and hierarchical signal bias removal (HSBR). Experimental results on various connected digits databases show a reduction in the digit error rate by 16%, 21%, and 28% when employing CMS, SBR, and HSBR, respectively. Our results also demonstrate that the HSBR technique outperforms SBR and CMS on every sub-data collection and exhibits consistent improvements even for short utterances.

Stochastic matching for robust speech recognition

Presents an approach to decrease the acoustic mismatch between a test utterance Y and a given set of speech hidden Markov models /spl Lambda//sub X/ to reduce the recognition performance degradation caused by possible distortions in the test utterance. This is accomplished by a parametric function that transforms either U or /spl Lambda//sub X/ to better match each other. The functional form of the transformation depends on prior knowledge about the mismatch, and the parameters are estimated along with the recognized string in a maximum-likelihood manner. experimental results verify the efficacy of the approach in improving the performance of a continuous speech recognition system in the presence of mismatch due to different transducers and transmission channels. >

A robust speaker-independent isolated word HMM recognizer for operation over the telephone network

This paper presents the results of a speaker-independent, isolated word speech recognition system developed for information access over Australian public switched telephone network (PSTN). The recognition system is based on Continuous Density Hidden Markov Modelling (CDHMM). The speech database was collected over the PSTN from a large variety of speakers and different geographical locations. The database contained a vocabulary of 55 words consisting of 41 country names and their variations plus a few control words. The recognition performance, tested on 100 other speakers (50 males and 50 females) with no grammar constraint, resulted in an overall recognition rate of 97.3%. This paper describes the HMM training methodology, which consisted of three stages: hand segmented seed model training, automatic word segmentation and reestimation. To facilitate the future implementation of the recognition system in a DSP environment, a fast frame synchronous Viterbi algorithm was implemented with no degradation in recognition performance. The end-point detection is performed by the combination of the silence/noise model with the word models. For confusable word pairs, sub-word models are used to improve the recognition rate. A post-processing approach is used to enhance the performance of the recognition system, in which all ranked candidates from the Viterbi decoding are subject to the tests of the minimum word duration and the likelihood difference between the first candidate and the second candidate.

Experiments in the application of isolated-word recognition to secondary driving controls for the disabled

Adaptive driving controls such as mechanical hand controls or electromechanical contact switches are now available which allow even the most severely impaired to drive. The residual functions, however, are all directed toward primary controls such as steering, braking, and accelerating, limiting the ability to operate secondary controls such as the horn, turn signals, ignition, and headlights, etc. In this paper, we discuss the application of speech recognition technology when operating these secondary controls. The performance of a speech recognition system inside a vehicle is studied, and the types of noise that degrade the recognition accuracy are also identified. Results are presented on the degradation in recognition performance caused by engine noise, fan noise, and interfering speech.

Measures of phonological rule set coverage

Some speech recognition systems use alternative lexical forms to cover various pronunciations encountered when people speak. However, excess forms in the lexicon degrade recognition performance. Thus lexicons should maximize coverage of actual pronunciations that occur, while minimizing the number of alternative pronunciations. A set of phonological rules for English as been developed at SRI to cover a large body of transcriptions of sentences read by American speakers. This phonological rule set covers 98.5% of the segments in a separate set of transcriptions. Rule efficiency can be defined as the ratio of the number of times forms generated by a rule were spoken in a corpus of test data to the number of forms generated by the rule for the sentences in that test data. By this definition, palatalization of /t/ and /d/ before tautosyllabic /r/ is efficient, while flapping of /t/ that follows /l/ is not. There is a trade‐off between the coverage of a rule set and the number of alternative pronunciations generated by the rule set. Rule efficiency can be used to rank rules in a rule set, to derive a family of rule sets that maximize coverage for a given number of alternative forms represented, or to select among alternative rule sets. [Work supported by DARPA.]