The AFsp package is a library of routines for reading and writing
audio files. The emphasis is on providing support for the type of
audio file used by the speech processing research community. The
routines have been designed to be easy to use, yet provide transparent
support the reading of several audio file formats. A secondary
purpose for distributing these routines is to encourage the use of a
standard audio file format for the header information in the output
files.
The following file formats are supported for reading.
- NIST SPHERE audio files
- Sun/NeXT audio files
- DEC audio files
- IRCAM SoundFiles
- INRS-Telecom audio files
- ESPS sampled data feature files
- Headerless audio files
The audio file open routine automatically senses the file type and
communicates it to the audio file reading routines. Formats are
converted on the fly as the file is read, so the user manipulates
floats and doesn’t need to worry about the underlying data format.
For writing, the routines produce a standard format file, though
options are available to produce headerless files if desired. This
standard format is a compatible with the Sun audio file format. There
is provision for storing extra information in the extensible part of
the header.
Several audio file utilities (for copying, comparing, and filtering
audio files) are included in the package.
www.cs.cmu.edu/afs/cs/project/ai-repository/ai/areas/speech/systems/afsp/0
The Audio File 