Recently various techniques to improve the correlation model of feature vector elements in speech recognition systems have been proposed. Such techniques include semi-tied covariance HMMs and systems based on factor analysis. All these schemes have been shown to improve the speech recognition performance without dramatically increasing the number of model parameters compared to standard diagonal covariance Gaussian mixture HMMs. This paper introduces a general form of acoustic model, the factor analysed HMM. A variety of configurations of this model and parameter sharing schemes, some of which correspond to standard systems, were examined. An EM algorithm for the parameter optimisation is presented along with a number of methods to increase the e#ciency of training. The performance of FAHMMs on medium to large vocabulary continuous speech recognition tasks was investigated. The experiments show that without elaborate complexity control an equivalent or better performance compared to a standard diagonal covariance Gaussian mixture HMM system can be achieved with considerably fewer parameters.

This paper examines the training of multiple-cluster systems using adaptive training schemes. Various forms of transformation and canonical model are described in a consistent framework allowing re-estimation formulae for all cases to be simply derived. Initial experiments using these various schemes on a large vocabulary speech recognition task are presented. The initial experiments indicate that to achieve best performance when adapting these multiple-cluster systems requires the use of adaptive training schemes rather than using simpler cluster initialisation schemes.

In automatic speech recognition complex state spaces are searched during the recognition process. By limiting these search spaces the computation time can be reduced, but unfortunately the recognition rate mostly decreases, too. However, especially for time-critical recognition tasks a search-space pruning is necessary. Therefore, we developed a dynamic mechanism to optimize the pruning parameters for time-constrained recognition tasks, e.g. speech recognition for robotic systems, in respect to word accuracy and computation time. With this mechanism an automatic speech recognition system can process speech signals with an approximately constant processing rate. Compared to a system without such a dynamic mechanism and the same time available for computation, the variance of the processing rate is decreased greatly without a significant loss of word accuracy. Furthermore, the extended system can be sped up to real-time processing, if desired or necessary.

Most modern automatic speech recognition systems make use of acoustic models based on hidden Markov models. To obtain reasonable recognition performance within a large vocabulary framework, the acoustic models usually include a pronunciation model, together with complex parameter tying schemes. In many cases the pronunciation model operates on a phoneme level and is derived independently of the underlying models. In contrast, this work is aimed at improving pronunciation modelling on a sub-phone level in a combined framework. The modelling of pronunciation variation is assumed to be of special importance for recognition of spontaneous speech.

Vocal Tract Length Normalisation (VTLN) is a commonly used technique to normalise for inter-speaker variability. It is based on the speaker-specific warping of the frequency axis, parameterised by a scalar warp factor. This factor is typically estimated using maximum likelihood. We discuss how VTLN may be applied to multiparty conversations, reporting a substantial decrease in word error rate in experiments using the ICSI meetings corpus. We investigate the behaviour of the VTLN warping factor and show that a stable estimate is not obtained. Instead it appears to be influenced by the context of the meeting, in particular the current conversational partner. These results are consistent with predictions made by the psycholinguistic interactive alignment account of dialogue, when applied at the acoustic and phonological levels.

Abstract—In this paper we investigate the combination of complementary acoustic feature streams in large vocabulary continuous speech recognition (LVCSR). We have explored the use of acoustic features obtained using a pitch-synchronous analysis, STRAIGHT, in combination with conventional features such as mel frequency cepstral coefficients. Pitch-synchronous acoustic features are of particular interest when used with vocal tract length normalisation (VTLN) which is known to be affected by the fundamental frequency. We have combined these spectral representations directly at the acoustic feature level using heteroscedastic linear discriminant analysis (HLDA) and at the system level using ROVER. We evaluated this approach on three LVCSR tasks: dictated newspaper text (WSJCAM0), conversational telephone speech (CTS), and multiparty meeting transcription. The CTS and meeting transcription experiments were both evaluated using standard NIST test sets and evaluation protocols. Our results indicate that combining conventional and pitch-synchronous acoustic feature sets using HLDA results in a consistent, significant decrease in word error rate across all three tasks. Combining at the system level using ROVER resulted in a further significant decrease in word error rate.

This paper presents a novel technique that allows testing spoken dialogue systems by means of an automatic generation of conversations. The technique permits to easily test spoken dialogue systems under a variety of lab-simulated conditions, as it is easy to vary or change the utterance corpus used to check the performance of the system. The technique is based on the use of a module called user simulator whose purpose is to behave as real users when they interact with dialogue systems. The behaviour of the simulator is decided by means of diverse scenarios that represent the goals of the users. The simulator aim is to achieve the goals set in the scenarios during the interaction with the dialogue system. We have applied the technique to test a dialogue system developed in our lab. The test has been carried out considering different levels of white and babble noise as well as a VTS noise compensation technique. The results prove that the dialogue system performance is worse under the babble noise conditions. The VTS technique has been effective when dealing with noisy utterances and has lead to better experimental results, particularly for the white noise. The technique has permitted to detect problems in the dialogue strategies employed to handle confirmation turns and recognition errors, suggesting that these strategies must be improved. q 2002 Elsevier Science B.V. All rights reserved.

Summary Most modern speech recognition systems use either Mel-frequency cepstral coefficients or per-ceptual linear prediction as acoustic features. Recently, there has been some interest in alter-native speech parameterisations based on using formant features. Formants are the resonant frequencies in the vocal tract which form the characteristic shape of the speech spectrum. How-ever, formants are difficult to reliably and robustly estimate from the speech signal and in some cases may not be clearly present. Rather than estimating the resonant frequencies, formant-like features can be used instead. Formant-like features use the characteristics of the spectral peaks to represent the spectrum. In this work, novel features are developed based on estimating a Gaussian mixture model (GMM) from the speech spectrum. This approach has previously been used sucessfully as a speech codec. The EM algorithm is used to estimate the parameters of the GMM. The extracted parameters: the means, standard deviations and component weights can be related to the for-mant locations, bandwidths and magnitudes. As the features directly represent the linear spec-trum, it is possibly to apply techniques for vocal tract length normalisation and additive noise

This paper presents a novel technique that allows testing spoken dialogue systems by means of an automatic generation of conversations. The technique permits to easily test spoken dialogue systems under a variety of lab-simulated conditions, as it is easy to vary or change the utterance corpus used to check the performance of the system. The technique is based on the use of a module called user simulator whose purpose is to behave as real users when they interact with dialogue systems. The behaviour of the simulator is decided by means of diverse scenarios that represent the goals of the users. The simulator aim is to achieve the goals set in the scenarios during the interaction with the dialogue system. We have applied the technique to test a dialogue system developed in our lab. The test has been carried out considering different levels of white and babble noise as well as a VTS noise compensation technique. The results prove that the dialogue system performance is worse under the babble noise conditions. The VTS technique has been effective when dealing with noisy utterances and has lead to better experimental results, particularly for the white noise. The technique has permitted to detect problems in the dialogue strategies employed to handle confirmation turns and recognition errors, suggesting that these strategies must be improved. q 2002 Elsevier Science B.V. All rights reserved.

We have investigated the use of a pitch adaptive spectral representation on large vocabulary speech recognition, in conjunction with speaker normalisation techniques. We have compared the effect of a smoothed spectrogram to the pitch adaptive spectral analysis by decoupling these two components of STRAIGHT. Experiments performed on a large vocabulary meeting speech recognition task highlight the importance of combining a pitch adaptive spectral representation with a conventional fixed window spectral analysis. We found evidence that STRAIGHT pitch adaptive features are more speaker independent than conventional MFCCs without pitch adaptation, thus they also provide better performances when combined using feature combination techniques such as Heteroscedastic