http://hdl.handle.net/10061/7704 Thu, 20 Jun 2013 04:51:16 GMT 2013-06-20T04:51:16Z http://hdl.handle.net/10061/8620 Title: Development of a toolkit handling multiple speech-oriented guidance agents for mobile applications Authors: Sunao Hara; Hiromichi Kawanami; Hiroshi Saruwatari; Kiyohiro Shikano Abstract: In this study, we propose a toolkit to handle multiple speech-oriented guidance agents for mobile applications. The basic architecture of the toolkit is server-and-client architecture. We assumed the servers are located on a cloud-computing environment, and the clients are mobile phones, such as the iPhone. It is difficult to develop an omnipotent spoken dialog system, but it is easy to develop a spoken dialog agent that has limited but deep knowledge. If such limited agents could communicate with each other, a spoken dialog system with wide-ranging knowledge could be created. Description: IWSDS2012: The 4th International Workshop on Spoken Dialog Systems, November 28-30, 2012, Paris, France Sun, 01 Jan 2012 00:00:00 GMT http://hdl.handle.net/10061/8620 2012-01-01T00:00:00Z http://hdl.handle.net/10061/8615 Title: Causal analysis of task completion erros in spoken music retrieval interactions Authors: Sunao Hara; Norihide Kitaoka; Kazuya Takeda Abstract: In this paper, we analyze the causes of task completion errors in spoken dialog systems, using a decision tree with N-gram features of the dialog to detect task-incomplete dialogs. The dialog for a music retrieval task is described by a sequence of tags related to user and system utterances and behaviors. The dialogs are manually classified into two classes: completed and uncompleted music retrieval tasks. Differences in tag classification performance between the two classes are discussed. We then construct decision trees which can detect if a dialog finished with the task completed or not, using information gain criterion. Decision trees using N-grams of manual tags and automatic tags achieved 74.2% and 80.4% classification accuracy, respectively, while the tree using interaction parameters achieved an accuracy rate of 65.7%. We also discuss more details of the causality of task incompletion for spoken dialog systems using such trees. Description: LREC2012: The 8th International Conference on Language Resources and Evaluation, May 21-27, 2012, Istanbul Sun, 01 Jan 2012 00:00:00 GMT http://hdl.handle.net/10061/8615 2012-01-01T00:00:00Z http://hdl.handle.net/10061/8616 Title: Object-based stereo up-mixer for wave field synthesis based on spatial information clustering Authors: Noriyoshi Kamado; Masayuki Hirata; Hiroshi Saruwatari; Kiyohiro Shikano Abstract: To build an acoustic system that can maintain the localization of sound images included in stereo mixed signals, we propose a new object-based up-mixer that performs sound source separation and sound location estimation. First, in a preliminary experiment, we show the effectiveness of sound location estimation using the proposed up-mixer via objective tests. Next, we evaluate the perception accuracy of sound localization by wave field synthesis using the proposed up-mixer via subjective tests. The results show that the proposed up-mixer provides a good localization of sound images included in stereo mixed signals at several listening positions. Description: EUSIPCO2012: The 20th European Signal Processing Conference, August 27-31, 2012, Bucharest, Romania Sun, 01 Jan 2012 00:00:00 GMT http://hdl.handle.net/10061/8616 2012-01-01T00:00:00Z http://hdl.handle.net/10061/8617 Title: Sound-localization-preserved binaural MMSE STSA estimator with explicit and implicit binaural cues Authors: Hiroshi Saruwatari; Ryo Wakisaka; Kiyohiro Shikano; Frederic Mustiere; Louis Thibault; Hossein Najaf-Zadeh; Martin Bouchard Abstract: In this paper, we address some variations of the sourcelocalization-preserved MMSE STSA estimator used for binaural hearing aids. In our previous work, the soundlocalization-preserved MMSE STSA estimator with ICAbased noise estimation has been proposed. However, this conventional method is based on an approximated optimization criterion and does not use binaural cues, resulting in poor noise reduction performance. To solve this problem, we propose two methods: a multichannel MMSE STSA estimator with explicit binaural cues, and a sound-localizationpreserved generalized MMSE STSA estimator with different speech priors for the left and right channels as implicit binaural cues. From the results of objective and subjective evaluation, we confirm that the noise reduction performance is improved using the proposed method. Description: EUSIPCO2012: The 20th European Signal Processing Conference, August 27-31, 2012, Bucharest, Romania Sun, 01 Jan 2012 00:00:00 GMT http://hdl.handle.net/10061/8617 2012-01-01T00:00:00Z