Robust Speech Recognition Method Research Articles

Robust Speech Recognition Using Teacher-Student Learning Domain Adaptation

Recently, robust speech recognition for real-world applications has attracted much attention. This paper proposes a robust speech recognition method based on the teacher-student learning framework for domain adaptation. In particular, the student network will be trained based on a novel optimization criterion defined by the encoder outputs of both teacher and student networks rather than the final output posterior probabilities, which aims to make the noisy audio map to the same embedding space as clean audio, so that the student network is adaptive in the noise domain. Comparative experiments demonstrate that the proposed method obtained good robustness against noise.

Environment mismatch compensation using average eigenspace-based methods for robust speech recognition

The performance of speech recognition systems is adversely affected by mismatch between training and test conditions due to environmental factors. In addition to the case of test data from noisy environments, there are scenarios where the training data itself is noisy. In this study, we propose a series of methods for mismatch compensation between training and test environments, based on our "average eigenspace" approach. These methods are also shown to be effective for non-stationary mismatch conditions. An advantage is that there is no need for explicit adaptation data since the method is applied to incoming test data to find the compensatory transform. We evaluate these approaches on two separate corpora which are collected from realistic car environments: CU-Move and UTDrive. Compared with a baseline system incorporating spectral subtraction, highpass filtering and cepstral mean normalization, we obtain a relative word error rate reduction of 17---26 % by applying the proposed techniques. These methods also result in a dimensionality reduction of the feature vectors allowing for a more compact set of acoustic models in the phoneme space, a property important for automatic speech recognition for small footprint size mobile devices such as cell-phone or PDA's which require ASR in diverse environments.

Improved modulation spectrum enhancement methods for robust speech recognition

In this paper, we present two novel algorithms to improve the noise robustness of features in speech recognition: modulation spectrum replacement (MSR) and modulation spectrum filtering (MSF). The magnitude spectra of feature streams are updated by referring to the information collected in the clean training set, and the resulting new feature streams are more noise-robust to achieve higher recognition accuracy. In experiments conducted on the Aurora-2 noisy digit database, we show that the proposed MSR achieves an average relative error reduction rate of nearly 57% compared to baseline processing, and MSF is specifically effective in enhancing the features preprocessed by conventional feature normalization methods to achieve even better recognition accuracy in noise-corrupted situations.

Noise Adaptive Training for Robust Automatic Speech Recognition

In traditional methods for noise robust automatic speech recognition, the acoustic models are typically trained using clean speech or using multi-condition data that is processed by the same feature enhancement algorithm expected to be used in decoding. In this paper, we propose a noise adaptive training (NAT) algorithm that can be applied to all training data that normalizes the environmental distortion as part of the model training. In contrast to feature enhancement methods, NAT estimates the underlying “pseudo-clean” model parameters directly without relying on point estimates of the clean speech features as an intermediate step. The pseudo-clean model parameters learned with NAT are later used with vector Taylor series (VTS) model adaptation for decoding noisy utterances at test time. Experiments performed on the Aurora 2 and Aurora 3 tasks demonstrate that the proposed NAT method obtain relative improvements of 18.83% and 32.02%, respectively, over VTS model adaptation.

Robust Speech Recognition by Combining Short-Term and Long-Term Spectrum Based Position-Dependent CMN with Conventional CMN

In a distant-talking environment, the length of channel impulse response is longer than the short-term spectral analysis window. Conventional short-term spectrum based Cepstral Mean Normalization (CMN) is therefore, not effective under these conditions. In this paper, we propose a robust speech recognition method by combining a short-term spectrum based CMN with a long-term one. We assume that a static speech segment (such as a vowel, for example) affected by reverberation, can be modeled by a long-term cepstral analysis. Thus, the effect of long reverberation on a static speech segment may be compensated by the long-term spectrum based CMN. The cepstral distance of neighboring frames is used to discriminate the static speech segment (long-term spectrum) and the non-static speech segment (short-term spectrum). The cepstra of the static and non-static speech segments are normalized by the corresponding cepstral means. In a previous study, we proposed an environmentally robust speech recognition method based on Position-Dependent CMN (PDCMN) to compensate for channel distortion depending on speaker position, and which is more efficient than conventional CMN. In this paper, the concept of combining short-term and long-term spectrum based CMN is extended to PDCMN. We call this Variable Term spectrum based PDCMN (VT-PDCMN). Since PDCMN/VT-PDCMN cannot normalize speaker variations because a position-dependent cepstral mean contains the average speaker characteristics over all speakers, we also combine PDCMN/VT-PDCMN with conventional CMN in this study. We conducted the experiments based on our proposed method using limited vocabulary (100 words) distant-talking isolated word recognition in a real environment. The proposed method achieved a relative error reduction rate of 60.9% over the conventional short-term spectrum based CMN and 30.6% over the short-term spectrum based PDCMN.

Effective Energy Feature Compensation Using Modified Log-energy Dynamic Range Normalization for Robust Speech Recognition

This paper proposes effective energy feature normalization methods for robust speech recognition in noisy environments. We first develop an energy subtraction method and a modified method for the Log-energy Dynamic Range Normalization (ERN) using inverse function. We then present the hybrid method combining the energy subtraction and the modified ERN. Using Aurora2.0 database for representative evaluations, a significant performance improvement over the ERN method is demonstrated.

The Performance Improvement of Speech Recognition System based on Stochastic Distance Measure

In this paper, we propose a robust speech recognition system under noisy environments. Since the presence of noise severely degrades the performance of speech recognition system, it is important to design the robust speech recognition method against noise. The proposed method adopts a new distance measure technique based on stochastic probability instead of conventional method using minimum error. For evaluating the performance of the proposed method, we compared it with conventional distance measure for the 10-isolated Korean digits with car noise. Here, the proposed method showed better recognition rate than conventional distance measure for the various car noisy environments.

A spatio-temporal speech enhancement scheme for robust speech recognition in noisy environments

A new speech enhancement scheme is presented integrating spatial and temporal signal processing methods for robust speech recognition in noisy environments. The scheme first separates spatially localized point sources from noisy speech signals recorded by two microphones. Blind source separation algorithms assuming no a priori knowledge about the sources involved are applied in this spatial processing stage. Then denoising of distributed background noise is achieved in a combined spatial/temporal processing approach. The desired speaker signal is first processed along with an artificially constructed noise signal in a supplementary blind source separation step. It is further denoised by exploiting differences in temporal speech and noise statistics in a wavelet filterbank. The scheme’s performance is illustrated by speech recognition experiments on real recordings in a noisy car environment. In comparison to a common multi-microphone technique like beamforming with spectral subtraction, the scheme is shown to enable more accurate speech recognition in the presence of a highly interfering point source and strong background noise.

Robust speech recognition method based on discriminative environment feature extraction

It is an effective approach to learn the influence of environmental parameters, such as additive noise and channel distortions, from training data for robust speech recognition. Most of the previous methods are based on maximum likelihood estimation criterion. However, these methods do not lead to a minimum error rate result. In this paper, a novel discriminative learning method of environmental parameters, which is based on Minimum Classification Error (MCE) criterion, is proposed. In the method, a simple classifier and the Generalized Probabilistic Descent (GPD) algorithm are adopted to iteratively learn the envirnnmental parameters. Consequently, the clean speech features are estimated from the noisy speech features with the estimated environmental parameters, and then the estimations of clean speech features are utilized in the back-end HMM classifier. Experiments show that the best error rate reduction of 32.1% is obtained, tested on a task of 18 isolated confusion Korean words, relative to aconventional HMM system.

A robust training algorithm for adverse speech recognition

In this paper, a new robust training algorithm is proposed for the generation of a set of bias-removed, noise-suppressed reference speech HMM models in adverse environment suffering from both channel bias and additive noise. Its main idea is to incorporate a signal bias-compensation operation and a PMC noise-compensation operation into its iterative training process. This makes the resulting speech HMM models more suitable to the given robust speech recognition method using the same signal bias-compensation and PMC noise-compensation operations in the recognition process. Experimental results showed that the speech HMM models it generated outperformed both the clean-speech HMM models and those generated by the conventional k-means algorithm for two adverse Mandarin speech recognition tasks. So it is a promising robust training algorithm.