Pub Date : 2013-10-01DOI: 10.1109/TASL.2013.2266794
C. Joder, S. Essid, G. Richard
This paper addresses the design of feature functions for the matching of a musical recording to the symbolic representation of the piece (the score). These feature functions are defined as dissimilarity measures between the audio observations and template vectors corresponding to the score. By expressing the template construction as a linear mapping from the symbolic to the audio representation, one can learn the feature functions by optimizing the linear transformation. In this paper, we explore two different learning strategies. The first one uses a best-fit criterion (minimum divergence), while the second one exploits a discriminative framework based on a Conditional Random Fields model (maximum likelihood criterion). We evaluate the influence of the feature functions in an audio-to-score alignment task, on a large database of popular and classical polyphonic music. The results show that with several types of models, using different temporal constraints, the learned mappings have the potential to outperform the classic heuristic mappings. Several representations of the audio observations, along with several distance functions are compared in this alignment task. Our experiments elect the symmetric Kullback-Leibler divergence. Moreover, both the spectrogram and a CQT-based representation turn out to provide very accurate alignments, detecting more than 97% of the onsets with a precision of 100 ms with our most complex system.
{"title":"Learning Optimal Features for Polyphonic Audio-to-Score Alignment","authors":"C. Joder, S. Essid, G. Richard","doi":"10.1109/TASL.2013.2266794","DOIUrl":"https://doi.org/10.1109/TASL.2013.2266794","url":null,"abstract":"This paper addresses the design of feature functions for the matching of a musical recording to the symbolic representation of the piece (the score). These feature functions are defined as dissimilarity measures between the audio observations and template vectors corresponding to the score. By expressing the template construction as a linear mapping from the symbolic to the audio representation, one can learn the feature functions by optimizing the linear transformation. In this paper, we explore two different learning strategies. The first one uses a best-fit criterion (minimum divergence), while the second one exploits a discriminative framework based on a Conditional Random Fields model (maximum likelihood criterion). We evaluate the influence of the feature functions in an audio-to-score alignment task, on a large database of popular and classical polyphonic music. The results show that with several types of models, using different temporal constraints, the learned mappings have the potential to outperform the classic heuristic mappings. Several representations of the audio observations, along with several distance functions are compared in this alignment task. Our experiments elect the symmetric Kullback-Leibler divergence. Moreover, both the spectrogram and a CQT-based representation turn out to provide very accurate alignments, detecting more than 97% of the onsets with a precision of 100 ms with our most complex system.","PeriodicalId":55014,"journal":{"name":"IEEE Transactions on Audio Speech and Language Processing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TASL.2013.2266794","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62891143","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-10-01DOI: 10.1109/TASL.2013.2266772
G. Degottex, Y. Stylianou
Voice models often use frequency limits to split the speech spectrum into two or more voiced/unvoiced frequency bands. However, from the voice production, the amplitude spectrum of the voiced source decreases smoothly without any abrupt frequency limit. Accordingly, multiband models struggle to estimate these limits and, as a consequence, artifacts can degrade the perceived quality. Using a linear frequency basis adapted to the non-stationarities of the speech signal, the Fan Chirp Transformation (FChT) have demonstrated harmonicity at frequencies higher than usually observed from the DFT which motivates a full-band modeling. The previously proposed Adaptive Quasi-Harmonic model (aQHM) offers even more flexibility than the FChT by using a non-linear frequency basis. In the current paper, exploiting the properties of aQHM, we describe a full-band Adaptive Harmonic Model (aHM) along with detailed descriptions of its corresponding algorithms for the estimation of harmonics up to the Nyquist frequency. Formal listening tests show that the speech reconstructed using aHM is nearly indistinguishable from the original speech. Experiments with synthetic signals also show that the proposed aHM globally outperforms previous sinusoidal and harmonic models in terms of precision in estimating the sinusoidal parameters. As a perspective, such a precision is interesting for building higher level models upon the sinusoidal parameters, like spectral envelopes for speech synthesis.
{"title":"Analysis and Synthesis of Speech Using an Adaptive Full-Band Harmonic Model","authors":"G. Degottex, Y. Stylianou","doi":"10.1109/TASL.2013.2266772","DOIUrl":"https://doi.org/10.1109/TASL.2013.2266772","url":null,"abstract":"Voice models often use frequency limits to split the speech spectrum into two or more voiced/unvoiced frequency bands. However, from the voice production, the amplitude spectrum of the voiced source decreases smoothly without any abrupt frequency limit. Accordingly, multiband models struggle to estimate these limits and, as a consequence, artifacts can degrade the perceived quality. Using a linear frequency basis adapted to the non-stationarities of the speech signal, the Fan Chirp Transformation (FChT) have demonstrated harmonicity at frequencies higher than usually observed from the DFT which motivates a full-band modeling. The previously proposed Adaptive Quasi-Harmonic model (aQHM) offers even more flexibility than the FChT by using a non-linear frequency basis. In the current paper, exploiting the properties of aQHM, we describe a full-band Adaptive Harmonic Model (aHM) along with detailed descriptions of its corresponding algorithms for the estimation of harmonics up to the Nyquist frequency. Formal listening tests show that the speech reconstructed using aHM is nearly indistinguishable from the original speech. Experiments with synthetic signals also show that the proposed aHM globally outperforms previous sinusoidal and harmonic models in terms of precision in estimating the sinusoidal parameters. As a perspective, such a precision is interesting for building higher level models upon the sinusoidal parameters, like spectral envelopes for speech synthesis.","PeriodicalId":55014,"journal":{"name":"IEEE Transactions on Audio Speech and Language Processing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TASL.2013.2266772","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62890382","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-10-01DOI: 10.1109/TASL.2013.2269291
Zhenhua Ling, L. Deng, Dong Yu
This paper presents a new spectral modeling method for statistical parametric speech synthesis. In the conventional methods, high-level spectral parameters, such as mel-cepstra or line spectral pairs, are adopted as the features for hidden Markov model (HMM)-based parametric speech synthesis. Our proposed method described in this paper improves the conventional method in two ways. First, distributions of low-level, un-transformed spectral envelopes (extracted by the STRAIGHT vocoder) are used as the parameters for synthesis. Second, instead of using single Gaussian distribution, we adopt the graphical models with multiple hidden variables, including restricted Boltzmann machines (RBM) and deep belief networks (DBN), to represent the distribution of the low-level spectral envelopes at each HMM state. At the synthesis time, the spectral envelopes are predicted from the RBM-HMMs or the DBN-HMMs of the input sentence following the maximum output probability parameter generation criterion with the constraints of the dynamic features. A Gaussian approximation is applied to the marginal distribution of the visible stochastic variables in the RBM or DBN at each HMM state in order to achieve a closed-form solution to the parameter generation problem. Our experimental results show that both RBM-HMM and DBN-HMM are able to generate spectral envelope parameter sequences better than the conventional Gaussian-HMM with superior generalization capabilities and that DBN-HMM and RBM-HMM perform similarly due possibly to the use of Gaussian approximation. As a result, our proposed method can significantly alleviate the over-smoothing effect and improve the naturalness of the conventional HMM-based speech synthesis system using mel-cepstra.
{"title":"Modeling Spectral Envelopes Using Restricted Boltzmann Machines and Deep Belief Networks for Statistical Parametric Speech Synthesis","authors":"Zhenhua Ling, L. Deng, Dong Yu","doi":"10.1109/TASL.2013.2269291","DOIUrl":"https://doi.org/10.1109/TASL.2013.2269291","url":null,"abstract":"This paper presents a new spectral modeling method for statistical parametric speech synthesis. In the conventional methods, high-level spectral parameters, such as mel-cepstra or line spectral pairs, are adopted as the features for hidden Markov model (HMM)-based parametric speech synthesis. Our proposed method described in this paper improves the conventional method in two ways. First, distributions of low-level, un-transformed spectral envelopes (extracted by the STRAIGHT vocoder) are used as the parameters for synthesis. Second, instead of using single Gaussian distribution, we adopt the graphical models with multiple hidden variables, including restricted Boltzmann machines (RBM) and deep belief networks (DBN), to represent the distribution of the low-level spectral envelopes at each HMM state. At the synthesis time, the spectral envelopes are predicted from the RBM-HMMs or the DBN-HMMs of the input sentence following the maximum output probability parameter generation criterion with the constraints of the dynamic features. A Gaussian approximation is applied to the marginal distribution of the visible stochastic variables in the RBM or DBN at each HMM state in order to achieve a closed-form solution to the parameter generation problem. Our experimental results show that both RBM-HMM and DBN-HMM are able to generate spectral envelope parameter sequences better than the conventional Gaussian-HMM with superior generalization capabilities and that DBN-HMM and RBM-HMM perform similarly due possibly to the use of Gaussian approximation. As a result, our proposed method can significantly alleviate the over-smoothing effect and improve the naturalness of the conventional HMM-based speech synthesis system using mel-cepstra.","PeriodicalId":55014,"journal":{"name":"IEEE Transactions on Audio Speech and Language Processing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TASL.2013.2269291","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62890772","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-10-01DOI: 10.1109/TASL.2013.2272513
Takuya Yoshioka, T. Nakatani
This paper proposes an approach, called noise model transfer (NMT), for estimating the rapidly changing parameter values of a feature-domain noise model, which can be used to enhance feature vectors corrupted by highly nonstationary noise. Unlike conventional methods, the proposed approach can exploit both observed feature vectors, representing spectral envelopes, and other signal properties that are usually discarded during feature extraction but that are useful for separating nonstationary noise from speech. Specifically, we assume the availability of a noise power spectrum estimator that can capture rapid changes in noise characteristics by leveraging such signal properties. NMT determines the optimal transformation from the estimated noise power spectra into the feature-domain noise model parameter values in the sense of maximum likelihood. NMT is successfully applied to meeting speech recognition, where the main noise sources are competing talkers; and reverberant speech recognition, where the late reverberation is regarded as highly nonstationary additive noise.
{"title":"Noise Model Transfer: Novel Approach to Robustness Against Nonstationary Noise","authors":"Takuya Yoshioka, T. Nakatani","doi":"10.1109/TASL.2013.2272513","DOIUrl":"https://doi.org/10.1109/TASL.2013.2272513","url":null,"abstract":"This paper proposes an approach, called noise model transfer (NMT), for estimating the rapidly changing parameter values of a feature-domain noise model, which can be used to enhance feature vectors corrupted by highly nonstationary noise. Unlike conventional methods, the proposed approach can exploit both observed feature vectors, representing spectral envelopes, and other signal properties that are usually discarded during feature extraction but that are useful for separating nonstationary noise from speech. Specifically, we assume the availability of a noise power spectrum estimator that can capture rapid changes in noise characteristics by leveraging such signal properties. NMT determines the optimal transformation from the estimated noise power spectra into the feature-domain noise model parameter values in the sense of maximum likelihood. NMT is successfully applied to meeting speech recognition, where the main noise sources are competing talkers; and reverberant speech recognition, where the late reverberation is regarded as highly nonstationary additive noise.","PeriodicalId":55014,"journal":{"name":"IEEE Transactions on Audio Speech and Language Processing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TASL.2013.2272513","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62891398","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-10-01DOI: 10.1109/TASL.2013.2263803
Yow-Bang Wang, Shang-Wen Li, Lin-Shan Lee
Gabor features have been proposed for extracting spectro-temporal modulation information from speech signals, and have been shown to yield large improvements in recognition accuracy. We use a flexible Tandem system framework that integrates multi-stream information including Gabor, MFCC, and pitch features in various ways, by modeling either or both of the tone and phoneme variations in Mandarin speech recognition. We use either phonemes or tonal phonemes (tonemes) as either the target classes of MLP posterior estimation and/or the acoustic units of HMM recognition. The experiments yield a comprehensive analysis on the contributions to recognition accuracy made by either of the feature sets. We discuss their complementarities in tone, phoneme, and toneme classification. We show that Gabor features are better for recognition of vowels and unvoiced consonants, while MFCCs are better for voiced consonants. Also, Gabor features are capable of capturing changes in signals across time and frequency bands caused by Mandarin tone patterns, while pitch features further offer extra tonal information. This explains why the integration of Gabor, MFCC, and pitch features offers such significant improvements.
{"title":"An Experimental Analysis on Integrating Multi-Stream Spectro-Temporal, Cepstral and Pitch Information for Mandarin Speech Recognition","authors":"Yow-Bang Wang, Shang-Wen Li, Lin-Shan Lee","doi":"10.1109/TASL.2013.2263803","DOIUrl":"https://doi.org/10.1109/TASL.2013.2263803","url":null,"abstract":"Gabor features have been proposed for extracting spectro-temporal modulation information from speech signals, and have been shown to yield large improvements in recognition accuracy. We use a flexible Tandem system framework that integrates multi-stream information including Gabor, MFCC, and pitch features in various ways, by modeling either or both of the tone and phoneme variations in Mandarin speech recognition. We use either phonemes or tonal phonemes (tonemes) as either the target classes of MLP posterior estimation and/or the acoustic units of HMM recognition. The experiments yield a comprehensive analysis on the contributions to recognition accuracy made by either of the feature sets. We discuss their complementarities in tone, phoneme, and toneme classification. We show that Gabor features are better for recognition of vowels and unvoiced consonants, while MFCCs are better for voiced consonants. Also, Gabor features are capable of capturing changes in signals across time and frequency bands caused by Mandarin tone patterns, while pitch features further offer extra tonal information. This explains why the integration of Gabor, MFCC, and pitch features offers such significant improvements.","PeriodicalId":55014,"journal":{"name":"IEEE Transactions on Audio Speech and Language Processing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TASL.2013.2263803","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62889905","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-10-01DOI: 10.1109/TASL.2013.2265087
Chao Zhang, Yi Liu, Yunqing Xia, Xuan Wang, Chin-Hui Lee
In this paper, we propose a discriminative dynamic Gaussian mixture selection (DGMS) strategy to generate reliable accent-specific units (ASUs) for multi-accent speech recognition. Time-aligned phone recognition is used to generate the ASUs that model accent variations explicitly and accurately. DGMS reconstructs and adjusts a pre-trained set of hidden Markov model (HMM) state densities to build dynamic observation densities for each input speech frame. A discriminative minimum classification error criterion is adopted to optimize the sizes of the HMM state observation densities with a genetic algorithm (GA). To the author's knowledge, the discriminative optimization for DGMS accomplishes discriminative training of discrete variables that is first proposed. We found the proposed framework is able to cover more multi-accent changes, thus reduce some performance loss in pruned beam search, without increasing the model size of the original acoustic model set. Evaluation on three typical Chinese accents, Chuan, Yue and Wu, shows that our approach outperforms traditional acoustic model reconstruction techniques with a syllable error rate reduction of 8.0%, 5.5% and 5.0%, respectively, while maintaining a good performance on standard Putonghua speech.
{"title":"Reliable Accent-Specific Unit Generation With Discriminative Dynamic Gaussian Mixture Selection for Multi-Accent Chinese Speech Recognition","authors":"Chao Zhang, Yi Liu, Yunqing Xia, Xuan Wang, Chin-Hui Lee","doi":"10.1109/TASL.2013.2265087","DOIUrl":"https://doi.org/10.1109/TASL.2013.2265087","url":null,"abstract":"In this paper, we propose a discriminative dynamic Gaussian mixture selection (DGMS) strategy to generate reliable accent-specific units (ASUs) for multi-accent speech recognition. Time-aligned phone recognition is used to generate the ASUs that model accent variations explicitly and accurately. DGMS reconstructs and adjusts a pre-trained set of hidden Markov model (HMM) state densities to build dynamic observation densities for each input speech frame. A discriminative minimum classification error criterion is adopted to optimize the sizes of the HMM state observation densities with a genetic algorithm (GA). To the author's knowledge, the discriminative optimization for DGMS accomplishes discriminative training of discrete variables that is first proposed. We found the proposed framework is able to cover more multi-accent changes, thus reduce some performance loss in pruned beam search, without increasing the model size of the original acoustic model set. Evaluation on three typical Chinese accents, Chuan, Yue and Wu, shows that our approach outperforms traditional acoustic model reconstruction techniques with a syllable error rate reduction of 8.0%, 5.5% and 5.0%, respectively, while maintaining a good performance on standard Putonghua speech.","PeriodicalId":55014,"journal":{"name":"IEEE Transactions on Audio Speech and Language Processing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TASL.2013.2265087","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62890303","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-10-01DOI: 10.1109/TASL.2013.2270370
S. Siniscalchi, Jinyu Li, Chin-Hui Lee
Model adaptation techniques are an efficient way to reduce the mismatch that typically occurs between the training and test condition of any automatic speech recognition (ASR) system. This work addresses the problem of increased degradation in performance when moving from speaker-dependent (SD) to speaker-independent (SI) conditions for connectionist (or hybrid) hidden Markov model/artificial neural network (HMM/ANN) systems in the context of large vocabulary continuous speech recognition (LVCSR). Adapting hybrid HMM/ANN systems on a small amount of adaptation data has been proven to be a difficult task, and has been a limiting factor in the widespread deployment of hybrid techniques in operational ASR systems. Addressing the crucial issue of speaker adaptation (SA) for hybrid HMM/ANN system can thereby have a great impact on the connectionist paradigm, which will play a major role in the design of next-generation LVCSR considering the great success reported by deep neural networks - ANNs with many hidden layers that adopts the pre-training technique - on many speech tasks. Current adaptation techniques for ANNs based on injecting an adaptable linear transformation network connected to either the input, or the output layer are not effective especially with a small amount of adaptation data, e.g., a single adaptation utterance. In this paper, a novel solution is proposed to overcome those limits and make it robust to scarce adaptation resources. The key idea is to adapt the hidden activation functions rather than the network weights. The adoption of Hermitian activation functions makes this possible. Experimental results on an LVCSR task demonstrate the effectiveness of the proposed approach.
{"title":"Hermitian Polynomial for Speaker Adaptation of Connectionist Speech Recognition Systems","authors":"S. Siniscalchi, Jinyu Li, Chin-Hui Lee","doi":"10.1109/TASL.2013.2270370","DOIUrl":"https://doi.org/10.1109/TASL.2013.2270370","url":null,"abstract":"Model adaptation techniques are an efficient way to reduce the mismatch that typically occurs between the training and test condition of any automatic speech recognition (ASR) system. This work addresses the problem of increased degradation in performance when moving from speaker-dependent (SD) to speaker-independent (SI) conditions for connectionist (or hybrid) hidden Markov model/artificial neural network (HMM/ANN) systems in the context of large vocabulary continuous speech recognition (LVCSR). Adapting hybrid HMM/ANN systems on a small amount of adaptation data has been proven to be a difficult task, and has been a limiting factor in the widespread deployment of hybrid techniques in operational ASR systems. Addressing the crucial issue of speaker adaptation (SA) for hybrid HMM/ANN system can thereby have a great impact on the connectionist paradigm, which will play a major role in the design of next-generation LVCSR considering the great success reported by deep neural networks - ANNs with many hidden layers that adopts the pre-training technique - on many speech tasks. Current adaptation techniques for ANNs based on injecting an adaptable linear transformation network connected to either the input, or the output layer are not effective especially with a small amount of adaptation data, e.g., a single adaptation utterance. In this paper, a novel solution is proposed to overcome those limits and make it robust to scarce adaptation resources. The key idea is to adapt the hidden activation functions rather than the network weights. The adoption of Hermitian activation functions makes this possible. Experimental results on an LVCSR task demonstrate the effectiveness of the proposed approach.","PeriodicalId":55014,"journal":{"name":"IEEE Transactions on Audio Speech and Language Processing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TASL.2013.2270370","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62891042","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-10-01DOI: 10.1109/TASL.2013.2270406
N. Gaubitch, M. Brookes, P. Naylor
We present an algorithm for blind estimation of the magnitude response of an acoustic channel from single microphone observations of a speech signal. The algorithm employs channel robust RASTA filtered Mel-frequency cepstral coefficients as features to train a Gaussian mixture model based classifier and average clean speech spectra are associated with each mixture; these are then used to blindly estimate the acoustic channel magnitude response from speech that has undergone spectral modification due to the channel. Experimental results using a variety of simulated and measured acoustic channels and additive babble noise, car noise and white Gaussian noise are presented. The results demonstrate that the proposed method is able to estimate a variety of channel magnitude responses to within an Itakura distance of dI ≤0.5 for SNR ≥10 dB.
{"title":"Blind Channel Magnitude Response Estimation in Speech Using Spectrum Classification","authors":"N. Gaubitch, M. Brookes, P. Naylor","doi":"10.1109/TASL.2013.2270406","DOIUrl":"https://doi.org/10.1109/TASL.2013.2270406","url":null,"abstract":"We present an algorithm for blind estimation of the magnitude response of an acoustic channel from single microphone observations of a speech signal. The algorithm employs channel robust RASTA filtered Mel-frequency cepstral coefficients as features to train a Gaussian mixture model based classifier and average clean speech spectra are associated with each mixture; these are then used to blindly estimate the acoustic channel magnitude response from speech that has undergone spectral modification due to the channel. Experimental results using a variety of simulated and measured acoustic channels and additive babble noise, car noise and white Gaussian noise are presented. The results demonstrate that the proposed method is able to estimate a variety of channel magnitude responses to within an Itakura distance of dI ≤0.5 for SNR ≥10 dB.","PeriodicalId":55014,"journal":{"name":"IEEE Transactions on Audio Speech and Language Processing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TASL.2013.2270406","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62891111","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-09-01DOI: 10.1109/TASL.2013.2261814
Muhammad Salman Khan, S. M. Naqvi, Ata ur-Rehman, Wenwu Wang, J. Chambers
Source separation algorithms that utilize only audio data can perform poorly if multiple sources or reverberation are present. In this paper we therefore propose a video-aided model-based source separation algorithm for a two-channel reverberant recording in which the sources are assumed static. By exploiting cues from video, we first localize individual speech sources in the enclosure and then estimate their directions. The interaural spatial cues, the interaural phase difference and the interaural level difference, as well as the mixing vectors are probabilistically modeled. The models make use of the source direction information and are evaluated at discrete time-frequency points. The model parameters are refined with the well-known expectation-maximization (EM) algorithm. The algorithm outputs time-frequency masks that are used to reconstruct the individual sources. Simulation results show that by utilizing the visual modality the proposed algorithm can produce better time-frequency masks thereby giving improved source estimates. We provide experimental results to test the proposed algorithm in different scenarios and provide comparisons with both other audio-only and audio-visual algorithms and achieve improved performance both on synthetic and real data. We also include dereverberation based pre-processing in our algorithm in order to suppress the late reverberant components from the observed stereo mixture and further enhance the overall output of the algorithm. This advantage makes our algorithm a suitable candidate for use in under-determined highly reverberant settings where the performance of other audio-only and audio-visual methods is limited.
{"title":"Video-Aided Model-Based Source Separation in Real Reverberant Rooms","authors":"Muhammad Salman Khan, S. M. Naqvi, Ata ur-Rehman, Wenwu Wang, J. Chambers","doi":"10.1109/TASL.2013.2261814","DOIUrl":"https://doi.org/10.1109/TASL.2013.2261814","url":null,"abstract":"Source separation algorithms that utilize only audio data can perform poorly if multiple sources or reverberation are present. In this paper we therefore propose a video-aided model-based source separation algorithm for a two-channel reverberant recording in which the sources are assumed static. By exploiting cues from video, we first localize individual speech sources in the enclosure and then estimate their directions. The interaural spatial cues, the interaural phase difference and the interaural level difference, as well as the mixing vectors are probabilistically modeled. The models make use of the source direction information and are evaluated at discrete time-frequency points. The model parameters are refined with the well-known expectation-maximization (EM) algorithm. The algorithm outputs time-frequency masks that are used to reconstruct the individual sources. Simulation results show that by utilizing the visual modality the proposed algorithm can produce better time-frequency masks thereby giving improved source estimates. We provide experimental results to test the proposed algorithm in different scenarios and provide comparisons with both other audio-only and audio-visual algorithms and achieve improved performance both on synthetic and real data. We also include dereverberation based pre-processing in our algorithm in order to suppress the late reverberant components from the observed stereo mixture and further enhance the overall output of the algorithm. This advantage makes our algorithm a suitable candidate for use in under-determined highly reverberant settings where the performance of other audio-only and audio-visual methods is limited.","PeriodicalId":55014,"journal":{"name":"IEEE Transactions on Audio Speech and Language Processing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TASL.2013.2261814","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62889329","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-09-01DOI: 10.1109/TASL.2013.2261813
R. Scharrer, M. Vorländer
The quality and performance of many multi-channel signal processing strategies in microphone arrays as well as mobile devices for the enhancement of speech intelligibility and audio quality depends to a large extent on the acoustic sound field that they are exposed to. As long as the assumption on the sound field is not met, the performance decreases significantly and may even yield worse results for the user than an unprocessed signal. Current hearing aids provide the user for instance with different programs to adapt the signal processing to the acoustic situation. Signal classification describes the signal content and not the type of sound field. Therefore, a further classification of the sound field, in addition to the signal classification, would increase the possibilities for an optimal adaption of the automatic program selection and the signal processing methods in mobile devices. To this end a sound field classification method is proposed that is based on the complex coherences between the input signals of distributed acoustic sensors. In addition to the general approach an exemplary setup of a hearing aid equipped with two microphone sensors is discussed. As only coherences are used, the method classifies the sound field regardless of the signal carried by it. This approach complements and extends the current signal classification approach used in common mobile devices. The method was successfully verified with simulated audio input signals and with real life examples.
{"title":"Sound Field Classification in Small Microphone Arrays Using Spatial Coherences","authors":"R. Scharrer, M. Vorländer","doi":"10.1109/TASL.2013.2261813","DOIUrl":"https://doi.org/10.1109/TASL.2013.2261813","url":null,"abstract":"The quality and performance of many multi-channel signal processing strategies in microphone arrays as well as mobile devices for the enhancement of speech intelligibility and audio quality depends to a large extent on the acoustic sound field that they are exposed to. As long as the assumption on the sound field is not met, the performance decreases significantly and may even yield worse results for the user than an unprocessed signal. Current hearing aids provide the user for instance with different programs to adapt the signal processing to the acoustic situation. Signal classification describes the signal content and not the type of sound field. Therefore, a further classification of the sound field, in addition to the signal classification, would increase the possibilities for an optimal adaption of the automatic program selection and the signal processing methods in mobile devices. To this end a sound field classification method is proposed that is based on the complex coherences between the input signals of distributed acoustic sensors. In addition to the general approach an exemplary setup of a hearing aid equipped with two microphone sensors is discussed. As only coherences are used, the method classifies the sound field regardless of the signal carried by it. This approach complements and extends the current signal classification approach used in common mobile devices. The method was successfully verified with simulated audio input signals and with real life examples.","PeriodicalId":55014,"journal":{"name":"IEEE Transactions on Audio Speech and Language Processing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TASL.2013.2261813","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62889264","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}