Pub Date : 2017-12-15DOI: 10.1109/APSIPA.2017.8282245
Rafael E. Banchs
Affect and emotion are inherent properties of human-human communication and interaction. Recent research interest in chatbots and conversational agents aims at making human-machine interaction more human-like in both behavioral and attitudinal terms. This paper intends to present some baby steps in this direction by analyzing a large dialogue dataset in terms of tonal, affective and emotional bias, with the objective of providing a valuable resource for developing and training datadriven conversational agents with discriminative power across such dimensions. Preliminary results of the conducted analysis demonstrate that only a relative small, although not negligible, percentage of the dialogue turns present clear orientation in any of the considered dimensions. Future research is still needed to determine whether this proportion is enough for biasing system responses in order to create different personality trends in conversational agents that are perceptible by humans when interacting with them.
{"title":"On the construction of more human-like chatbots: Affect and emotion analysis of movie dialogue data","authors":"Rafael E. Banchs","doi":"10.1109/APSIPA.2017.8282245","DOIUrl":"https://doi.org/10.1109/APSIPA.2017.8282245","url":null,"abstract":"Affect and emotion are inherent properties of human-human communication and interaction. Recent research interest in chatbots and conversational agents aims at making human-machine interaction more human-like in both behavioral and attitudinal terms. This paper intends to present some baby steps in this direction by analyzing a large dialogue dataset in terms of tonal, affective and emotional bias, with the objective of providing a valuable resource for developing and training datadriven conversational agents with discriminative power across such dimensions. Preliminary results of the conducted analysis demonstrate that only a relative small, although not negligible, percentage of the dialogue turns present clear orientation in any of the considered dimensions. Future research is still needed to determine whether this proportion is enough for biasing system responses in order to create different personality trends in conversational agents that are perceptible by humans when interacting with them.","PeriodicalId":142091,"journal":{"name":"2017 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-12-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128908679","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}
Serpentine locomotion is highly coordinating and full of adaptive ability in a clutter environment. Such outstanding and unique characteristics are acquired through millions of years' evolution. It is highly desirable to enhance robot with such characteristics, which is one of the ultimate aims of biomimetic research. To achieve this goal, we adopt a central pattern generator (CPG) inspired controller to generate Serpentine locomotion in a crawling robot. According to biology studies, CPGs are a set of neuronal circuits, which are responsible for producing rhythmic motion employed in animal locomotion. Such locomotion generation approach makes use of a set of coupled Kuramoto Oscillators to imitate of CPG in a nerve system. Moreover, to deal with dynamically changing environments, a feedback based on fuzzy logic control strategy is investigated. Finally, the proposed control approach is verified through the experiments of a crawling robot prototype.
{"title":"Locomotion control of a serpentine crawling robot inspired by central pattern generators","authors":"Jiadong Wang, Wenjuan Ouyang, Wenchao Gao, Qinyuan Ren","doi":"10.1109/APSIPA.2017.8282067","DOIUrl":"https://doi.org/10.1109/APSIPA.2017.8282067","url":null,"abstract":"Serpentine locomotion is highly coordinating and full of adaptive ability in a clutter environment. Such outstanding and unique characteristics are acquired through millions of years' evolution. It is highly desirable to enhance robot with such characteristics, which is one of the ultimate aims of biomimetic research. To achieve this goal, we adopt a central pattern generator (CPG) inspired controller to generate Serpentine locomotion in a crawling robot. According to biology studies, CPGs are a set of neuronal circuits, which are responsible for producing rhythmic motion employed in animal locomotion. Such locomotion generation approach makes use of a set of coupled Kuramoto Oscillators to imitate of CPG in a nerve system. Moreover, to deal with dynamically changing environments, a feedback based on fuzzy logic control strategy is investigated. Finally, the proposed control approach is verified through the experiments of a crawling robot prototype.","PeriodicalId":142091,"journal":{"name":"2017 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-12-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121317282","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 : 2017-12-15DOI: 10.1109/APSIPA.2017.8282038
M. Ramanathan, W. Yau, E. Teoh, N. Magnenat-Thalmann
Recognition of actions from videos is a difficult task due to several factors like dynamic backgrounds, occlusion, pose-variations observed. To tackle the pose variation problem, we propose a simple method based on a novel set of pose-invariant kinematic features which are encoded in a human body centric space. The proposed framework begins with detection of neck point, which will serve as a origin of body centric space. We propose a deep learning based classifier to detect neck point based on the output of fully connected network layer. With the help of the detected neck, propagation mechanism is proposed to divide the foreground region into head, torso and leg grids. The motion observed in each of these body part grids are represented using a set of pose-invariant kinematic features. These features represent motion of foreground or body region with respect to the detected neck point's motion and encoded based on view in a human body centric space. Based on these features, poseinvariant action recognition can be achieved. Due to the body centric space is used, non-upright human posture actions can also be handled easily. To test its effectiveness in non-upright human postures in actions, a new dataset is introduced with 8 non-upright actions performed by 35 subjects in 3 different views. Experiments have been conducted on benchmark and newly proposed non-upright action dataset to identify limitations and get insights on the proposed framework.
{"title":"Pose-invariant kinematic features for action recognition","authors":"M. Ramanathan, W. Yau, E. Teoh, N. Magnenat-Thalmann","doi":"10.1109/APSIPA.2017.8282038","DOIUrl":"https://doi.org/10.1109/APSIPA.2017.8282038","url":null,"abstract":"Recognition of actions from videos is a difficult task due to several factors like dynamic backgrounds, occlusion, pose-variations observed. To tackle the pose variation problem, we propose a simple method based on a novel set of pose-invariant kinematic features which are encoded in a human body centric space. The proposed framework begins with detection of neck point, which will serve as a origin of body centric space. We propose a deep learning based classifier to detect neck point based on the output of fully connected network layer. With the help of the detected neck, propagation mechanism is proposed to divide the foreground region into head, torso and leg grids. The motion observed in each of these body part grids are represented using a set of pose-invariant kinematic features. These features represent motion of foreground or body region with respect to the detected neck point's motion and encoded based on view in a human body centric space. Based on these features, poseinvariant action recognition can be achieved. Due to the body centric space is used, non-upright human posture actions can also be handled easily. To test its effectiveness in non-upright human postures in actions, a new dataset is introduced with 8 non-upright actions performed by 35 subjects in 3 different views. Experiments have been conducted on benchmark and newly proposed non-upright action dataset to identify limitations and get insights on the proposed framework.","PeriodicalId":142091,"journal":{"name":"2017 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-12-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133865510","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 : 2017-12-14DOI: 10.1109/APSIPA.2017.8282124
Jonghee Kim, Jinsu Kim, Seokeon Choi, Muhammad Abul Hasan, Changick Kim
In this paper, we propose a deep convolutional feature-based robust and efficient template matching method. The originality of the proposed method is that it is based on a scale-adaptive feature extraction approach. This approach is influenced by an observation that each layer in a CNN represents a different level of deep features of the actual image contents. In order to keep the features scalable, we extract deep feature vectors of the template and the input image adaptively from a layer of a CNN. By using such scalable and deep representation of the image contents, we attempt to solve the template matching by measuring the similarity between the features of the template and the input image using an efficient similarity measuring technique called normalized cross-correlation (NCC). Using NCC helps in avoiding redundant computations of adjacent patches caused by the sliding window approach. As a result, the proposed method achieves state-of-the-art template matching performance and lowers the computational cost significantly than the state-of- the-art methods in the literature.
{"title":"Robust template matching using scale-adaptive deep convolutional features","authors":"Jonghee Kim, Jinsu Kim, Seokeon Choi, Muhammad Abul Hasan, Changick Kim","doi":"10.1109/APSIPA.2017.8282124","DOIUrl":"https://doi.org/10.1109/APSIPA.2017.8282124","url":null,"abstract":"In this paper, we propose a deep convolutional feature-based robust and efficient template matching method. The originality of the proposed method is that it is based on a scale-adaptive feature extraction approach. This approach is influenced by an observation that each layer in a CNN represents a different level of deep features of the actual image contents. In order to keep the features scalable, we extract deep feature vectors of the template and the input image adaptively from a layer of a CNN. By using such scalable and deep representation of the image contents, we attempt to solve the template matching by measuring the similarity between the features of the template and the input image using an efficient similarity measuring technique called normalized cross-correlation (NCC). Using NCC helps in avoiding redundant computations of adjacent patches caused by the sliding window approach. As a result, the proposed method achieves state-of-the-art template matching performance and lowers the computational cost significantly than the state-of- the-art methods in the literature.","PeriodicalId":142091,"journal":{"name":"2017 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124954486","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 : 2017-12-14DOI: 10.1109/APSIPA.2017.8282220
Hyungjun Lim, Younggwan Kim, Yoonhoe Kim, Hoirin Kim
This paper addresses the problem of query-by-example spoken term detection (QbE-STD) in the presence of background noises that are inevitable in real applications. To deal with this, we propose a convolutional neural network (CNN) based bottleneck feature representation for a keyword. A combined network that is made by attaching a bottleneck layer on top of a CNN is trained on Wall Street Journal (WSJ) database. Finally, dynamic time warping (DTW) based template matching is performed to measure the distance between enrollment and test feature matrices which are extracted from the bottleneck layer. The proposed method is evaluated in terms of equal error rate (EER) on Aurora 4 Database. A series of experimental results verify that the proposed method performs significantly better than the baseline system in noisy environments shows over 30% relative equal error rate (EER) improvement in average.
{"title":"CNN-based bottleneck feature for noise robust query-by-example spoken term detection","authors":"Hyungjun Lim, Younggwan Kim, Yoonhoe Kim, Hoirin Kim","doi":"10.1109/APSIPA.2017.8282220","DOIUrl":"https://doi.org/10.1109/APSIPA.2017.8282220","url":null,"abstract":"This paper addresses the problem of query-by-example spoken term detection (QbE-STD) in the presence of background noises that are inevitable in real applications. To deal with this, we propose a convolutional neural network (CNN) based bottleneck feature representation for a keyword. A combined network that is made by attaching a bottleneck layer on top of a CNN is trained on Wall Street Journal (WSJ) database. Finally, dynamic time warping (DTW) based template matching is performed to measure the distance between enrollment and test feature matrices which are extracted from the bottleneck layer. The proposed method is evaluated in terms of equal error rate (EER) on Aurora 4 Database. A series of experimental results verify that the proposed method performs significantly better than the baseline system in noisy environments shows over 30% relative equal error rate (EER) improvement in average.","PeriodicalId":142091,"journal":{"name":"2017 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"305 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124347782","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 : 2017-12-12DOI: 10.1109/APSIPA.2017.8282065
Li Jun, Tee Keng Peng, C. Lawrence, Wan Kong Wah, Yau Wei Yun
Automatically delivering objects to in-car passengers has many potential applications. Such a system generally consists of two sub-systems: a perception system and an action system. The perception system basically looks for the targets' positions and the action system delivers objects to the targets. In this paper, we propose a novel perception system, which contains two major functions: estimation of reaching points and discovering potential risks. The reaching points are the locations where robot arms needs to reach. Moreover, it should be able to reach with comfort by passengers and keep a safe distance from the car body. In order to achieve this, all the vehicle components (side surfaces, side mirrors etc.), which may cause collision, need to be detected. Potential risks are usually caused by moving objects or changing door state (close to open) during the operation. It is necessary to monitor these two situations to avoid any potential risks during operation. Our offline test shows that the accuracy of reaching points estimation can reach up to 94% and the response time for moving objects detection or door state changes is less than 1 millisecond.
{"title":"A perception system for robot arms to convey objects to in-car passengers","authors":"Li Jun, Tee Keng Peng, C. Lawrence, Wan Kong Wah, Yau Wei Yun","doi":"10.1109/APSIPA.2017.8282065","DOIUrl":"https://doi.org/10.1109/APSIPA.2017.8282065","url":null,"abstract":"Automatically delivering objects to in-car passengers has many potential applications. Such a system generally consists of two sub-systems: a perception system and an action system. The perception system basically looks for the targets' positions and the action system delivers objects to the targets. In this paper, we propose a novel perception system, which contains two major functions: estimation of reaching points and discovering potential risks. The reaching points are the locations where robot arms needs to reach. Moreover, it should be able to reach with comfort by passengers and keep a safe distance from the car body. In order to achieve this, all the vehicle components (side surfaces, side mirrors etc.), which may cause collision, need to be detected. Potential risks are usually caused by moving objects or changing door state (close to open) during the operation. It is necessary to monitor these two situations to avoid any potential risks during operation. Our offline test shows that the accuracy of reaching points estimation can reach up to 94% and the response time for moving objects detection or door state changes is less than 1 millisecond.","PeriodicalId":142091,"journal":{"name":"2017 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"31 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116042461","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 : 2017-12-01DOI: 10.1109/APSIPA.2017.8282066
Yi Feng, Zhifeng Huang, Yun Zhang
In this paper, the motion planning of 6-Dofs robot arm for bandaging nursing task is proposed. With the increase in the degree of social aging, and inconvenient old man is easily injured in daily life, the robot which can handle the simple task of medical care can give them a great help. In this task, DH method is used to establish the robot model, and using the RPY coordinate transformation to plan the spiral trajectory. And then detect the collision between the joint or link of the robot and the injured place. Later with the result of the detection, changing the end effector's posture to avoid collision in order to avoid the damage to wound. By using the method a simulation experiment is made in MATLAB. And the collision-free trajectory motion is realized in the bandaging simulation experiment.
{"title":"Motion planning of a 6-Dofs robot arm for bandaging nursing task","authors":"Yi Feng, Zhifeng Huang, Yun Zhang","doi":"10.1109/APSIPA.2017.8282066","DOIUrl":"https://doi.org/10.1109/APSIPA.2017.8282066","url":null,"abstract":"In this paper, the motion planning of 6-Dofs robot arm for bandaging nursing task is proposed. With the increase in the degree of social aging, and inconvenient old man is easily injured in daily life, the robot which can handle the simple task of medical care can give them a great help. In this task, DH method is used to establish the robot model, and using the RPY coordinate transformation to plan the spiral trajectory. And then detect the collision between the joint or link of the robot and the injured place. Later with the result of the detection, changing the end effector's posture to avoid collision in order to avoid the damage to wound. By using the method a simulation experiment is made in MATLAB. And the collision-free trajectory motion is realized in the bandaging simulation experiment.","PeriodicalId":142091,"journal":{"name":"2017 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115227581","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 : 2017-12-01DOI: 10.1109/APSIPA.2017.8282277
Ryo Masumura, Taichi Asami, H. Masataki, Y. Aono
This paper reports an initial study of unsupervised adaptation that assumes simultaneous use of both n-gram and recurrent neural network (RNN) language models (LMs) in automatic speech recognition (ASR). It is known that a combination of n-grams and RNN LMs is a more effective approach to ASR than using each of them singly. However, unsupervised adaptation methods that simultaneously adapt both n-grams and RNN LMs have not been presented while various unsupervised adaptation methods specific to either n-gram LMs or RNN LMs have been examined. In order to handle different LMs in a unified unsupervised adaptation framework, our key idea is to introduce mixture modeling for both n-gram LMs and RNN LMs. The mixture modeling can simultaneously handle multiple LMs and unsupervised adaptation can be easily accomplished merely by adjusting their mixture weights using a recognition hypothesis of an input speech. This paper proposes joint unsupervised adaptation achieved by a hybrid mixture modeling using both n-gram mixture models and RNN mixture models. We present latent Dirichlet allocation based hybrid mixture modeling for effective topic adaptation. Our experiments in lecture ASR tasks show the effectiveness of joint unsupervised adaptation. We also reveal performance in which only one n-gram or RNN LM is adapted.
{"title":"Joint unsupervised adaptation of n-gram and RNN language models via LDA-based hybrid mixture modeling","authors":"Ryo Masumura, Taichi Asami, H. Masataki, Y. Aono","doi":"10.1109/APSIPA.2017.8282277","DOIUrl":"https://doi.org/10.1109/APSIPA.2017.8282277","url":null,"abstract":"This paper reports an initial study of unsupervised adaptation that assumes simultaneous use of both n-gram and recurrent neural network (RNN) language models (LMs) in automatic speech recognition (ASR). It is known that a combination of n-grams and RNN LMs is a more effective approach to ASR than using each of them singly. However, unsupervised adaptation methods that simultaneously adapt both n-grams and RNN LMs have not been presented while various unsupervised adaptation methods specific to either n-gram LMs or RNN LMs have been examined. In order to handle different LMs in a unified unsupervised adaptation framework, our key idea is to introduce mixture modeling for both n-gram LMs and RNN LMs. The mixture modeling can simultaneously handle multiple LMs and unsupervised adaptation can be easily accomplished merely by adjusting their mixture weights using a recognition hypothesis of an input speech. This paper proposes joint unsupervised adaptation achieved by a hybrid mixture modeling using both n-gram mixture models and RNN mixture models. We present latent Dirichlet allocation based hybrid mixture modeling for effective topic adaptation. Our experiments in lecture ASR tasks show the effectiveness of joint unsupervised adaptation. We also reveal performance in which only one n-gram or RNN LM is adapted.","PeriodicalId":142091,"journal":{"name":"2017 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"59 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115242580","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 : 2017-12-01DOI: 10.1109/APSIPA.2017.8282133
Ying Shi, Askar Hamdullah, Zhiyuan Tang, Dong Wang, T. Zheng
Automatic speech recognition (ASR) has gained significant improvement for major languages such as English and Chinese, partly due to the emergence of deep neural networks (DNN) and large amount of training data. For minority languages, however, the progress is largely behind the main stream. A particularly obstacle is that there are almost no large-scale speech databases for minority languages, and the only few databases are held by some institutes as private properties, far from open and standard, and very few are free. Besides the speech database, phonetic and linguistic resources are also scarce, including phone set, lexicon, and language model. In this paper, we publish a speech database in Kazakh, a major minority language in the western China. Accompanying this database, a full set of phonetic and linguistic resources are also published, by which a full-fledged Kazakh ASR system can be constructed. We will describe the recipe for constructing a baseline system, and report our present results. The resources are free for research institutes and can be obtained by request. The publication is supported by the M2ASR project supported by NSFC, which aims to build multilingual ASR systems for minority languages in China.
{"title":"A free Kazakh speech database and a speech recognition baseline","authors":"Ying Shi, Askar Hamdullah, Zhiyuan Tang, Dong Wang, T. Zheng","doi":"10.1109/APSIPA.2017.8282133","DOIUrl":"https://doi.org/10.1109/APSIPA.2017.8282133","url":null,"abstract":"Automatic speech recognition (ASR) has gained significant improvement for major languages such as English and Chinese, partly due to the emergence of deep neural networks (DNN) and large amount of training data. For minority languages, however, the progress is largely behind the main stream. A particularly obstacle is that there are almost no large-scale speech databases for minority languages, and the only few databases are held by some institutes as private properties, far from open and standard, and very few are free. Besides the speech database, phonetic and linguistic resources are also scarce, including phone set, lexicon, and language model. In this paper, we publish a speech database in Kazakh, a major minority language in the western China. Accompanying this database, a full set of phonetic and linguistic resources are also published, by which a full-fledged Kazakh ASR system can be constructed. We will describe the recipe for constructing a baseline system, and report our present results. The resources are free for research institutes and can be obtained by request. The publication is supported by the M2ASR project supported by NSFC, which aims to build multilingual ASR systems for minority languages in China.","PeriodicalId":142091,"journal":{"name":"2017 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115682427","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 : 2017-12-01DOI: 10.1109/APSIPA.2017.8282191
E. Cahyaningtyas, D. Arifianto
Although Bahasa Indonesia is used by about 263 milion people in the world, it is calssified into an under- resourced language. In this paper we outlined the development of casual sentences of Bahasa Indonesia speech corpus in which contains a speech database and its transcription. Firstly, we selected casual Bahasa Indonesia sentences from movie and drama trasncript and formed 1029 declarative sentences and 500 question sentences, respectively. We hired six professional radio news readers to utter the sentences to avoid local dialect in sound-proof booth. Then segmentation and labeling was performed to make create transcription including the time label of each invidual phoneme. To ensure the quality of the database, we manually inspected the waveform and the frequency of the individual sentences using spectrogram. The results suggest that the speech corpus may be used for speech processing project like speech recognition and speech synthesis. In the on-going research, we are developing high quality of speech synthesis, namely speaker adaptation and speaker averaging.
{"title":"Development of under-resourced Bahasa Indonesia speech corpus","authors":"E. Cahyaningtyas, D. Arifianto","doi":"10.1109/APSIPA.2017.8282191","DOIUrl":"https://doi.org/10.1109/APSIPA.2017.8282191","url":null,"abstract":"Although Bahasa Indonesia is used by about 263 milion people in the world, it is calssified into an under- resourced language. In this paper we outlined the development of casual sentences of Bahasa Indonesia speech corpus in which contains a speech database and its transcription. Firstly, we selected casual Bahasa Indonesia sentences from movie and drama trasncript and formed 1029 declarative sentences and 500 question sentences, respectively. We hired six professional radio news readers to utter the sentences to avoid local dialect in sound-proof booth. Then segmentation and labeling was performed to make create transcription including the time label of each invidual phoneme. To ensure the quality of the database, we manually inspected the waveform and the frequency of the individual sentences using spectrogram. The results suggest that the speech corpus may be used for speech processing project like speech recognition and speech synthesis. In the on-going research, we are developing high quality of speech synthesis, namely speaker adaptation and speaker averaging.","PeriodicalId":142091,"journal":{"name":"2017 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124435699","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：