Pub Date : 2013-12-01DOI: 10.1109/ASRU.2013.6707755
Yosuke Kashiwagi, D. Saito, N. Minematsu, K. Hirose
In this paper, we propose the use of deep neural networks to expand conventional methods of statistical feature enhancement based on piecewise linear transformation. Stereo-based piecewise linear compensation for environments (SPLICE), which is a powerful statistical approach for feature enhancement, models the probabilistic distribution of input noisy features as a mixture of Gaussians. However, soft assignment of an input vector to divided regions is sometimes done inadequately and the vector comes to go through inadequate conversion. Especially when conversion has to be linear, the conversion performance will be easily degraded. Feature enhancement using neural networks is another powerful approach which can directly model a non-linear relationship between noisy and clean feature spaces. In this case, however, it tends to suffer from over-fitting problems. In this paper, we attempt to mitigate this problem by reducing the number of model parameters to estimate. Our neural network is trained whose output layer is associated with the states in the clean feature space, not in the noisy feature space. This strategy makes the size of the output layer independent of the kind of a given noisy environment. Firstly, we characterize the distribution of clean features as a Gaussian mixture model and then, by using deep neural networks, estimate discriminatively the state in the clean space that an input noisy feature corresponds to. Experimental evaluations using the Aurora 2 dataset demonstrate that our proposed method has the best performance compared to conventional methods.
{"title":"Discriminative piecewise linear transformation based on deep learning for noise robust automatic speech recognition","authors":"Yosuke Kashiwagi, D. Saito, N. Minematsu, K. Hirose","doi":"10.1109/ASRU.2013.6707755","DOIUrl":"https://doi.org/10.1109/ASRU.2013.6707755","url":null,"abstract":"In this paper, we propose the use of deep neural networks to expand conventional methods of statistical feature enhancement based on piecewise linear transformation. Stereo-based piecewise linear compensation for environments (SPLICE), which is a powerful statistical approach for feature enhancement, models the probabilistic distribution of input noisy features as a mixture of Gaussians. However, soft assignment of an input vector to divided regions is sometimes done inadequately and the vector comes to go through inadequate conversion. Especially when conversion has to be linear, the conversion performance will be easily degraded. Feature enhancement using neural networks is another powerful approach which can directly model a non-linear relationship between noisy and clean feature spaces. In this case, however, it tends to suffer from over-fitting problems. In this paper, we attempt to mitigate this problem by reducing the number of model parameters to estimate. Our neural network is trained whose output layer is associated with the states in the clean feature space, not in the noisy feature space. This strategy makes the size of the output layer independent of the kind of a given noisy environment. Firstly, we characterize the distribution of clean features as a Gaussian mixture model and then, by using deep neural networks, estimate discriminatively the state in the clean space that an input noisy feature corresponds to. Experimental evaluations using the Aurora 2 dataset demonstrate that our proposed method has the best performance compared to conventional methods.","PeriodicalId":265258,"journal":{"name":"2013 IEEE Workshop on Automatic Speech Recognition and Understanding","volume":"341 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115673587","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 : 2013-12-01DOI: 10.1109/ASRU.2013.6707761
Oliver Walter, Timo Korthals, Reinhold Häb-Umbach, B. Raj
Discovering the linguistic structure of a language solely from spoken input asks for two steps: phonetic and lexical discovery. The first is concerned with identifying the categorical subword unit inventory and relating it to the underlying acoustics, while the second aims at discovering words as repeated patterns of subword units. The hierarchical approach presented here accounts for classification errors in the first stage by modelling the pronunciation of a word in terms of subword units probabilistically: a hidden Markov model with discrete emission probabilities, emitting the observed subword unit sequences. We describe how the system can be learned in a completely unsupervised fashion from spoken input. To improve the initialization of the training of the word pronunciations, the output of a dynamic time warping based acoustic pattern discovery system is used, as it is able to discover similar temporal sequences in the input data. This improved initialization, using only weak supervision, has led to a 40% reduction in word error rate on a digit recognition task.
{"title":"A hierarchical system for word discovery exploiting DTW-based initialization","authors":"Oliver Walter, Timo Korthals, Reinhold Häb-Umbach, B. Raj","doi":"10.1109/ASRU.2013.6707761","DOIUrl":"https://doi.org/10.1109/ASRU.2013.6707761","url":null,"abstract":"Discovering the linguistic structure of a language solely from spoken input asks for two steps: phonetic and lexical discovery. The first is concerned with identifying the categorical subword unit inventory and relating it to the underlying acoustics, while the second aims at discovering words as repeated patterns of subword units. The hierarchical approach presented here accounts for classification errors in the first stage by modelling the pronunciation of a word in terms of subword units probabilistically: a hidden Markov model with discrete emission probabilities, emitting the observed subword unit sequences. We describe how the system can be learned in a completely unsupervised fashion from spoken input. To improve the initialization of the training of the word pronunciations, the output of a dynamic time warping based acoustic pattern discovery system is used, as it is able to discover similar temporal sequences in the input data. This improved initialization, using only weak supervision, has led to a 40% reduction in word error rate on a digit recognition task.","PeriodicalId":265258,"journal":{"name":"2013 IEEE Workshop on Automatic Speech Recognition and Understanding","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130532728","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 : 2013-12-01DOI: 10.1109/ASRU.2013.6707706
Udhyakumar Nallasamy, Mark C. Fuhs, M. Woszczyna, Florian Metze, Tanja Schultz
Speaker dependent (SD) ASR systems have significantly lower word error rates (WER) compared to speaker independent (SI) systems. However, SD systems require sufficient training data from the target speaker, which is impractical to collect in a short time. We present a technique for training SD models using just few minutes of speaker's data. We compensate for the lack of adequate speaker-specific data by selecting neighbours from a database of existing speakers who are acoustically close to the target speaker. These neighbours provide ample training data, which is used to adapt the SI model to obtain an initial SD model for the new speaker with significantly lower WER. We evaluate various neighbour selection algorithms on a large-scale medical transcription task and report significant reduction in WER using only 5 mins of speaker-specific data. We conduct a detailed analysis of various factors such as gender and accent in the neighbour selection. Finally, we study neighbour selection and adaptation in the context of discriminative objective functions.
{"title":"Neighbour selection and adaptation for rapid speaker-dependent ASR","authors":"Udhyakumar Nallasamy, Mark C. Fuhs, M. Woszczyna, Florian Metze, Tanja Schultz","doi":"10.1109/ASRU.2013.6707706","DOIUrl":"https://doi.org/10.1109/ASRU.2013.6707706","url":null,"abstract":"Speaker dependent (SD) ASR systems have significantly lower word error rates (WER) compared to speaker independent (SI) systems. However, SD systems require sufficient training data from the target speaker, which is impractical to collect in a short time. We present a technique for training SD models using just few minutes of speaker's data. We compensate for the lack of adequate speaker-specific data by selecting neighbours from a database of existing speakers who are acoustically close to the target speaker. These neighbours provide ample training data, which is used to adapt the SI model to obtain an initial SD model for the new speaker with significantly lower WER. We evaluate various neighbour selection algorithms on a large-scale medical transcription task and report significant reduction in WER using only 5 mins of speaker-specific data. We conduct a detailed analysis of various factors such as gender and accent in the neighbour selection. Finally, we study neighbour selection and adaptation in the context of discriminative objective functions.","PeriodicalId":265258,"journal":{"name":"2013 IEEE Workshop on Automatic Speech Recognition and Understanding","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129717422","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 : 2013-12-01DOI: 10.1109/ASRU.2013.6707734
A. Ivanov, S. Jalalvand, R. Gretter, D. Falavigna
We explore the impact of speech- and speaker-specific modeling onto the Modulation Spectrum Analysis - Kolmogorov-Smirnov feature Testing (MSA-KST) characterization method in the task of automated prediction of the cognitive impairment diagnosis, namely dysphasia and pervasive development disorder. Phoneme-synchronous capturing of speech dynamics is a reasonable choice for a segmental speech characterization system as it allows comparing speech dynamics in the similar phonetic contexts. Speaker-specific modeling aims at reducing the “within-the-class” variability of the characterized speech or speaker population by removing the effect of speaker properties that should have no relation to the characterization. Specifically the vocal tract length of a speaker has nothing to do with the diagnosis attribution and, thus, the feature set shall be normalized accordingly. The resulting system compares favorably to the baseline system of the Interspeech'2013 Computational Paralinguistics Challenge.
{"title":"Phonetic and anthropometric conditioning of MSA-KST cognitive impairment characterization system","authors":"A. Ivanov, S. Jalalvand, R. Gretter, D. Falavigna","doi":"10.1109/ASRU.2013.6707734","DOIUrl":"https://doi.org/10.1109/ASRU.2013.6707734","url":null,"abstract":"We explore the impact of speech- and speaker-specific modeling onto the Modulation Spectrum Analysis - Kolmogorov-Smirnov feature Testing (MSA-KST) characterization method in the task of automated prediction of the cognitive impairment diagnosis, namely dysphasia and pervasive development disorder. Phoneme-synchronous capturing of speech dynamics is a reasonable choice for a segmental speech characterization system as it allows comparing speech dynamics in the similar phonetic contexts. Speaker-specific modeling aims at reducing the “within-the-class” variability of the characterized speech or speaker population by removing the effect of speaker properties that should have no relation to the characterization. Specifically the vocal tract length of a speaker has nothing to do with the diagnosis attribution and, thus, the feature set shall be normalized accordingly. The resulting system compares favorably to the baseline system of the Interspeech'2013 Computational Paralinguistics Challenge.","PeriodicalId":265258,"journal":{"name":"2013 IEEE Workshop on Automatic Speech Recognition and Understanding","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125468980","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 : 2013-12-01DOI: 10.1109/ASRU.2013.6707732
Duc Le, E. Provost
Research in emotion recognition seeks to develop insights into the temporal properties of emotion. However, automatic emotion recognition from spontaneous speech is challenging due to non-ideal recording conditions and highly ambiguous ground truth labels. Further, emotion recognition systems typically work with noisy high-dimensional data, rendering it difficult to find representative features and train an effective classifier. We tackle this problem by using Deep Belief Networks, which can model complex and non-linear high-level relationships between low-level features. We propose and evaluate a suite of hybrid classifiers based on Hidden Markov Models and Deep Belief Networks. We achieve state-of-the-art results on FAU Aibo, a benchmark dataset in emotion recognition [1]. Our work provides insights into important similarities and differences between speech and emotion.
{"title":"Emotion recognition from spontaneous speech using Hidden Markov models with deep belief networks","authors":"Duc Le, E. Provost","doi":"10.1109/ASRU.2013.6707732","DOIUrl":"https://doi.org/10.1109/ASRU.2013.6707732","url":null,"abstract":"Research in emotion recognition seeks to develop insights into the temporal properties of emotion. However, automatic emotion recognition from spontaneous speech is challenging due to non-ideal recording conditions and highly ambiguous ground truth labels. Further, emotion recognition systems typically work with noisy high-dimensional data, rendering it difficult to find representative features and train an effective classifier. We tackle this problem by using Deep Belief Networks, which can model complex and non-linear high-level relationships between low-level features. We propose and evaluate a suite of hybrid classifiers based on Hidden Markov Models and Deep Belief Networks. We achieve state-of-the-art results on FAU Aibo, a benchmark dataset in emotion recognition [1]. Our work provides insights into important similarities and differences between speech and emotion.","PeriodicalId":265258,"journal":{"name":"2013 IEEE Workshop on Automatic Speech Recognition and Understanding","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126311824","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 : 2013-12-01DOI: 10.1109/ASRU.2013.6707738
A. Touazi, M. Debyeche
This paper proposes a new scheme for low bit-rate source coding of Mel Frequency Cepstral Coefficients (MFCCs) in Distributed Speech Recognition (DSR) system. The method uses the compressed ETSI Advanced Front-End (ETSI-AFE) features factorized into SVD components. By investigating the correlation property between successive MFCC frames, the odd ones are encoded using ETSI-AFE, while only the singular values and the nearest left singular vectors index are encoded and transmitted for the even frames. At the server side, the non-transmitted MFCCs are evaluated through their quantized singular values and the nearest left singular vectors. The system provides a compression bit-rate of 2.7 kbps. The recognition experiments were carried out on the Aurora-2 database for clean and multi-condition training modes. The simulation results show good recognition performance without significant degradation, with respect to the ETSI-AFE encoder.
{"title":"An SVD-based scheme for MFCC compression in distributed speech recognition system","authors":"A. Touazi, M. Debyeche","doi":"10.1109/ASRU.2013.6707738","DOIUrl":"https://doi.org/10.1109/ASRU.2013.6707738","url":null,"abstract":"This paper proposes a new scheme for low bit-rate source coding of Mel Frequency Cepstral Coefficients (MFCCs) in Distributed Speech Recognition (DSR) system. The method uses the compressed ETSI Advanced Front-End (ETSI-AFE) features factorized into SVD components. By investigating the correlation property between successive MFCC frames, the odd ones are encoded using ETSI-AFE, while only the singular values and the nearest left singular vectors index are encoded and transmitted for the even frames. At the server side, the non-transmitted MFCCs are evaluated through their quantized singular values and the nearest left singular vectors. The system provides a compression bit-rate of 2.7 kbps. The recognition experiments were carried out on the Aurora-2 database for clean and multi-condition training modes. The simulation results show good recognition performance without significant degradation, with respect to the ETSI-AFE encoder.","PeriodicalId":265258,"journal":{"name":"2013 IEEE Workshop on Automatic Speech Recognition and Understanding","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126447524","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 : 2013-12-01DOI: 10.1109/ASRU.2013.6707703
Yuuki Tachioka, Shinji Watanabe, Jonathan Le Roux, J. Hershey
This paper proposes a generalized discriminative training framework for system combination, which encompasses acoustic modeling (Gaussian mixture models and deep neural networks) and discriminative feature transformation. To improve the performance by combining base systems with complementary systems, complementary systems should have reasonably good performance while tending to have different outputs compared with the base system. Although it is difficult to balance these two somewhat opposite targets in conventional heuristic combination approaches, our framework provides a new objective function that enables to adjust the balance within a sequential discriminative training criterion. We also describe how the proposed method relates to boosting methods. Experiments on highly noisy middle vocabulary speech recognition task (2nd CHiME challenge track 2) and LVCSR task (Corpus of Spontaneous Japanese) show the effectiveness of the proposed method, compared with a conventional system combination approach.
{"title":"A generalized discriminative training framework for system combination","authors":"Yuuki Tachioka, Shinji Watanabe, Jonathan Le Roux, J. Hershey","doi":"10.1109/ASRU.2013.6707703","DOIUrl":"https://doi.org/10.1109/ASRU.2013.6707703","url":null,"abstract":"This paper proposes a generalized discriminative training framework for system combination, which encompasses acoustic modeling (Gaussian mixture models and deep neural networks) and discriminative feature transformation. To improve the performance by combining base systems with complementary systems, complementary systems should have reasonably good performance while tending to have different outputs compared with the base system. Although it is difficult to balance these two somewhat opposite targets in conventional heuristic combination approaches, our framework provides a new objective function that enables to adjust the balance within a sequential discriminative training criterion. We also describe how the proposed method relates to boosting methods. Experiments on highly noisy middle vocabulary speech recognition task (2nd CHiME challenge track 2) and LVCSR task (Corpus of Spontaneous Japanese) show the effectiveness of the proposed method, compared with a conventional system combination approach.","PeriodicalId":265258,"journal":{"name":"2013 IEEE Workshop on Automatic Speech Recognition and Understanding","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121489053","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 : 2013-12-01DOI: 10.1109/ASRU.2013.6707714
Emmanuel Ferreira, F. Lefèvre
This paper investigates the conditions under which expert knowledge can be used to accelerate the policy optimization of a learning agent. Recent works on reinforcement learning for dialogue management allowed to devise sophisticated methods for value estimation in order to deal all together with exploration/exploitation dilemma, sample-efficiency and non-stationary environments. In this paper, a reward shaping method and an exploration scheme, both based on some intuitive hand-coded expert advices, are combined with an efficient temporal difference-based learning procedure. The key objective is to boost the initial training stage, when the system is not sufficiently reliable to interact with real users (e.g. clients). Our claims are illustrated by experiments based on simulation and carried out using a state-of-the-art goal-oriented dialogue management framework, the Hidden Information State (HIS).
{"title":"Expert-based reward shaping and exploration scheme for boosting policy learning of dialogue management","authors":"Emmanuel Ferreira, F. Lefèvre","doi":"10.1109/ASRU.2013.6707714","DOIUrl":"https://doi.org/10.1109/ASRU.2013.6707714","url":null,"abstract":"This paper investigates the conditions under which expert knowledge can be used to accelerate the policy optimization of a learning agent. Recent works on reinforcement learning for dialogue management allowed to devise sophisticated methods for value estimation in order to deal all together with exploration/exploitation dilemma, sample-efficiency and non-stationary environments. In this paper, a reward shaping method and an exploration scheme, both based on some intuitive hand-coded expert advices, are combined with an efficient temporal difference-based learning procedure. The key objective is to boost the initial training stage, when the system is not sufficiently reliable to interact with real users (e.g. clients). Our claims are illustrated by experiments based on simulation and carried out using a state-of-the-art goal-oriented dialogue management framework, the Hidden Information State (HIS).","PeriodicalId":265258,"journal":{"name":"2013 IEEE Workshop on Automatic Speech Recognition and Understanding","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121518288","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 : 2013-09-05DOI: 10.1109/ASRU.2013.6707747
Tara N. Sainath, L. Horesh, Brian Kingsbury, A. Aravkin, B. Ramabhadran
Hessian-free training has become a popular parallel second order optimization technique for Deep Neural Network training. This study aims at speeding up Hessian-free training, both by means of decreasing the amount of data used for training, as well as through reduction of the number of Krylov subspace solver iterations used for implicit estimation of the Hessian. In this paper, we develop an L-BFGS based preconditioning scheme that avoids the need to access the Hessian explicitly. Since L-BFGS cannot be regarded as a fixed-point iteration, we further propose the employment of flexible Krylov subspace solvers that retain the desired theoretical convergence guarantees of their conventional counterparts. Second, we propose a new sampling algorithm, which geometrically increases the amount of data utilized for gradient and Krylov subspace iteration calculations. On a 50-hr English Broadcast News task, we find that these methodologies provide roughly a 1.5× speed-up, whereas, on a 300-hr Switchboard task, these techniques provide over a 2.3× speedup, with no loss in WER. These results suggest that even further speed-up is expected, as problems scale and complexity grows.
{"title":"Accelerating Hessian-free optimization for Deep Neural Networks by implicit preconditioning and sampling","authors":"Tara N. Sainath, L. Horesh, Brian Kingsbury, A. Aravkin, B. Ramabhadran","doi":"10.1109/ASRU.2013.6707747","DOIUrl":"https://doi.org/10.1109/ASRU.2013.6707747","url":null,"abstract":"Hessian-free training has become a popular parallel second order optimization technique for Deep Neural Network training. This study aims at speeding up Hessian-free training, both by means of decreasing the amount of data used for training, as well as through reduction of the number of Krylov subspace solver iterations used for implicit estimation of the Hessian. In this paper, we develop an L-BFGS based preconditioning scheme that avoids the need to access the Hessian explicitly. Since L-BFGS cannot be regarded as a fixed-point iteration, we further propose the employment of flexible Krylov subspace solvers that retain the desired theoretical convergence guarantees of their conventional counterparts. Second, we propose a new sampling algorithm, which geometrically increases the amount of data utilized for gradient and Krylov subspace iteration calculations. On a 50-hr English Broadcast News task, we find that these methodologies provide roughly a 1.5× speed-up, whereas, on a 300-hr Switchboard task, these techniques provide over a 2.3× speedup, with no loss in WER. These results suggest that even further speed-up is expected, as problems scale and complexity grows.","PeriodicalId":265258,"journal":{"name":"2013 IEEE Workshop on Automatic Speech Recognition and Understanding","volume":"73 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117244156","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 : 2013-09-05DOI: 10.1109/ASRU.2013.6707749
Tara N. Sainath, Brian Kingsbury, Abdel-rahman Mohamed, George E. Dahl, G. Saon, H. Soltau, T. Beran, A. Aravkin, B. Ramabhadran
Deep Convolutional Neural Networks (CNNs) are more powerful than Deep Neural Networks (DNN), as they are able to better reduce spectral variation in the input signal. This has also been confirmed experimentally, with CNNs showing improvements in word error rate (WER) between 4-12% relative compared to DNNs across a variety of LVCSR tasks. In this paper, we describe different methods to further improve CNN performance. First, we conduct a deep analysis comparing limited weight sharing and full weight sharing with state-of-the-art features. Second, we apply various pooling strategies that have shown improvements in computer vision to an LVCSR speech task. Third, we introduce a method to effectively incorporate speaker adaptation, namely fMLLR, into log-mel features. Fourth, we introduce an effective strategy to use dropout during Hessian-free sequence training. We find that with these improvements, particularly with fMLLR and dropout, we are able to achieve an additional 2-3% relative improvement in WER on a 50-hour Broadcast News task over our previous best CNN baseline. On a larger 400-hour BN task, we find an additional 4-5% relative improvement over our previous best CNN baseline.
{"title":"Improvements to Deep Convolutional Neural Networks for LVCSR","authors":"Tara N. Sainath, Brian Kingsbury, Abdel-rahman Mohamed, George E. Dahl, G. Saon, H. Soltau, T. Beran, A. Aravkin, B. Ramabhadran","doi":"10.1109/ASRU.2013.6707749","DOIUrl":"https://doi.org/10.1109/ASRU.2013.6707749","url":null,"abstract":"Deep Convolutional Neural Networks (CNNs) are more powerful than Deep Neural Networks (DNN), as they are able to better reduce spectral variation in the input signal. This has also been confirmed experimentally, with CNNs showing improvements in word error rate (WER) between 4-12% relative compared to DNNs across a variety of LVCSR tasks. In this paper, we describe different methods to further improve CNN performance. First, we conduct a deep analysis comparing limited weight sharing and full weight sharing with state-of-the-art features. Second, we apply various pooling strategies that have shown improvements in computer vision to an LVCSR speech task. Third, we introduce a method to effectively incorporate speaker adaptation, namely fMLLR, into log-mel features. Fourth, we introduce an effective strategy to use dropout during Hessian-free sequence training. We find that with these improvements, particularly with fMLLR and dropout, we are able to achieve an additional 2-3% relative improvement in WER on a 50-hour Broadcast News task over our previous best CNN baseline. On a larger 400-hour BN task, we find an additional 4-5% relative improvement over our previous best CNN baseline.","PeriodicalId":265258,"journal":{"name":"2013 IEEE Workshop on Automatic Speech Recognition and Understanding","volume":"53 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127609422","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：