Telecommunications & Signal Processing Laboratory
AFsp Programs and Routines
The AFsp package provides audio file utility programs and a library of
routines for reading and writing audio files.
Audio File Utilities:
- Compare audio files, producing statistics and signal-to-noise ratio
- Copy audio files. This program combines samples from input audio files (an
arbitrary linear combination) and writes them to the output file in a user
selectable format. One application is to provide format conversion for an
audio file; another is to combine or select samples from multi-channel files.
- Filter audio files. This program filters an audio file with an FIR, IIR or
all-pole filter. This program can be used with an appropriate filter for
sample rate conversion. The AFsp package includes a number of filters
from the ITU-T Software Tool Library (G.191) in a form that can be used
directly with the FiltAudio program.
- Generate noise. This program generates an audio file containing Gaussian
- Generate a sinusoid. This program generates an audio file containing a
sine wave of a given amplitude and frequency.
- Display information about an audio file.
- Demonstration program: LPC analysis on a speech file.
- Demonstration program: LPC synthesis from a residual file.
- Perceptual evaluation of audio quality.
- Resample data from an audio file. This process involves interpolating
between the samples in the original file to create a new sequence of samples
with a new spacing (sampling rate).
Audio File Library Routines:
The audio file open routine automatically senses the input file type. The
data format for the file is passed to the audio file reading routines. The audio
file reading routine does byte swapping and format conversion on the fly as the
file is read. The user sees floating-point data without needing to worry about
the underlying data format.
Once a file is opened for writing, the user calls routines which take the
user's data and do byte swapping and format conversion of the data
The library routines supplied in AFsp package form part of the libtsp
signal processing library. These routines are written in C.
- Close an audio file
- Read data from an audio file (return double values)
- Write data to an audio file (double input values)
- Read data from an audio file (return float values)
- Write data to an audio file (float input values)
- Open an audio file for reading
- Open an audio file for writing
- Explicitly set the input audio file type
- Set the audio file information string
- Set defaults for headerless audio files
- Set defaults for loudspeaker spatial positions
Audio File Formats:
The following file formats are supported for reading.
- Headerless audio files
- AU audio files
- WAVE files
- AIFF/AIFF-C audio files
- NIST SPHERE audio files
- IRCAM soundfiles
- INRS-Telecom audio files
- CSL NSP files
- ESPS sampled data feature files
- Comdisco SPW Signal files
- Text audio files
The following file formats are supported for writing.
- Headerless audio files (including text data)
- AU audio files
- WAVE files
- AIFF or AIFF-C audio files
Audio File Data Types:
The following (uncompressed) data types are supported.
- 8-bit µ-law and A-law
- 8-bit signed and unsigned fixed-point
- 9-bit to 32-bit signed fixed-point
- 32-bit floating-point
- 64-bit floating-point
- text data