Presentation | 2005/6/16 A Study of subword models for vocabulary-free spoken document retrieval system and its application for Retrieving TV Broadcasts in a Disaster Kohei IWATA, Yoshiaki ITOH, Kazunori KOJIMA, Masaaki ISHIGAME, Kazuyo TANAKA, Shi wook Lee, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | According to the recent spread of personal computers and video hard-disc recorders, a new function is needed such that it is easy for a user to identify the scene that a user wants to watch in a video data. For this purpose, this paper proposes a speech retrieval method that does not use a general speech recognizer but a subword models. The method is characterized by using subword models and acoustic distance between the subword models. We conducted some experiments for evaluating the retrieval performance between triphone that are general models in a speech recognizer, demi-phone and Sub-Phonetic Segment (SPS) that are more precise models than triphone models on time axis, and illustrated a better performance was obtained by using new models such as SPS and acoustic distances between subword models. Furthermore, we applied the method to the application of retrieving the safety information in TV broadcast of Niigata-Chuetsu earthquake, and confirmed the necessity and possibility of the proposed system. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Speech retrieval / subword / distance between models / disaster information |
Paper # | SP2005-21 |
Date of Issue |
Conference Information | |
Committee | SP |
---|---|
Conference Date | 2005/6/16(1days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | |
Vice Chair | |
Secretary | |
Assistant |
Paper Information | |
Registration To | Speech (SP) |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | A Study of subword models for vocabulary-free spoken document retrieval system and its application for Retrieving TV Broadcasts in a Disaster |
Sub Title (in English) | |
Keyword(1) | Speech retrieval |
Keyword(2) | subword |
Keyword(3) | distance between models |
Keyword(4) | disaster information |
1st Author's Name | Kohei IWATA |
1st Author's Affiliation | Faculty of Software and Information Science Iwate Prefectural University() |
2nd Author's Name | Yoshiaki ITOH |
2nd Author's Affiliation | Faculty of Software and Information Science Iwate Prefectural University |
3rd Author's Name | Kazunori KOJIMA |
3rd Author's Affiliation | Faculty of Software and Information Science Iwate Prefectural University |
4th Author's Name | Masaaki ISHIGAME |
4th Author's Affiliation | Faculty of Software and Information Science Iwate Prefectural University |
5th Author's Name | Kazuyo TANAKA |
5th Author's Affiliation | Institute of Library and Information Science Tsukuba University |
6th Author's Name | Shi wook Lee |
6th Author's Affiliation | Information Technology Research Institute AIST |
Date | 2005/6/16 |
Paper # | SP2005-21 |
Volume (vol) | vol.105 |
Number (no) | 132 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |