First Experiences with a System for Content Based Retrieval of Information from Speech Recordings (1995)
| Venue: | Speech Recordings,” IJCAI-95 Workshop on Intelligent Multimedia Information Retrieval, M.Maybury (chair), working notes |
| Citations: | 23 - 0 self |
BibTeX
@INPROCEEDINGS{Schäuble95firstexperiences,
author = {Peter Schäuble and Martin Wechsler},
title = {First Experiences with a System for Content Based Retrieval of Information from Speech Recordings},
booktitle = {Speech Recordings,” IJCAI-95 Workshop on Intelligent Multimedia Information Retrieval, M.Maybury (chair), working notes},
year = {1995},
pages = {59--69}
}
Bookmark
 |
 |
 |
 |
 |
 |
OpenURL
Abstract
We present a speech retrieval system aimed at retrieving information from audio recordings containing speech. The current system contains 4.5 hours of radio news and accepts textual queries. The fully automatic indexing was done using speech recognition techniques. Indexing speech documents is challenging, because word boundaries are difficult to detect, and recognition errors influence the retrieval effectiveness. Our indexing vocabulary consists of 5000 phone sequences. To obtain such a set of suitable indexing features, we developed an efficient selection algorithm which uses statistical information derived from the audio collection. 1 Introduction We will demonstrate one of the first speech retrieval systems and we will report on our first experiences with this system. A speech retrieval system accepts vague queries and it performs a best-match search to find those speech recordings that are relevant to the query. The database of our current speech retrieval system consists of 0.5...
Citations
No citations identified.
