Abstract
In this paper, we investigate the use of a speaker identification technique to solve the barge-in speech detection problem. This scenario is a very simple application of speaker identification since only two users are involved. This is further simplified by the fact that the prompt speaker can be modelled apriori. Additionally, the user can be modelled as well improving the performance of the system on subsequent utterances. In the system described below, we explicitly model several non-speech sounds such as laughter, coughs and breath noises. We show that this technique is generally better than that of current methods which measure the ratio of incoming speech energy to that of the prompt signal being played.
Original language | English (US) |
---|---|
Pages | 327-330 |
Number of pages | 4 |
State | Published - 1999 |
Event | 6th European Conference on Speech Communication and Technology, EUROSPEECH 1999 - Budapest, Hungary Duration: Sep 5 1999 → Sep 9 1999 |
Conference
Conference | 6th European Conference on Speech Communication and Technology, EUROSPEECH 1999 |
---|---|
Country/Territory | Hungary |
City | Budapest |
Period | 9/5/99 → 9/9/99 |
All Science Journal Classification (ASJC) codes
- Computer Science Applications
- Software
- Linguistics and Language
- Communication