Accession Number:

ADA507788

Title:

Speech Music Discrimination Using Class-Specific Features

Descriptive Note:

Conference paper

Corporate Author:

NAVAL UNDERSEA WARFARE CENTER DIV NEWPORT RI

Personal Author(s):

• Beierholm, Thomas
• Baggenstoss, Paul M.

Report Date:

2004-08-01

Pagination or Media Count:

5.0

Abstract:

In this paper the application of the class-specific features approach to classification is demonstrated for the problem of discriminating between speech and music. Feature extraction is class-specific and can therefore be tailored to each class meaning that segment size, model orders and the type of features used can be different for the classes. The performance of the discriminator is evaluated and an example of how classification is possible without training is given.

Descriptors:

• *SPEECH
• *MUSIC
• *DISCRIMINATION
• CLASSIFICATION
• FEATURE EXTRACTION
• AUDITORY SIGNALS
• TRAINING
• SYMPOSIA
• ALGORITHMS

Subject Categories:

• Acoustics
• Voice Communications

Distribution Statement:

APPROVED FOR PUBLIC RELEASE