Home My Page Projects SPro
Summary Activity Tracker Lists SCM Files

Project Filelist for SPro

File Release Notes and Changelog

Release Name: 4.0.1

Release Notes

The Speech Signal Processing (SPro) toolkit provides the standard speech signal analysis routines. The basic functions needed to build signal processing algorithms are defined in a library and runtime tools are provided for the standard analysis procedure (LPC, filter-bank and cepstral analysis).


SPro is distributed "as is" under the GNU Public License.


 Library (libspro.a):

  - system.h: system dependent stuff
  - spro.h: all the definitions for libspro.a
  - sig.c: waveform I/O functions
  - header.c: feature stream header functions
  - spf.c: feature I/O functions
  - misc.c: miscellaneous stuff that didn't fit anywhere else
  - lpc.c: LPC based speech analysis functions
  - fft.c: FFT based speech analysis functions
  - convert.c: delta, delta-delta, CMS and other convertions

  - scopy: copy/list of an SPro data file into various formats (HTK, ascii, Sirocco, ...)
  - sfbank: filter bank analysis
  - sfbcep: filter bank cepstral coefficients
  - slpc: LPC analysis
  - slpcep: LPC-based cepstral coefficients


  - spro.texi


  See INSTALL for details.


  As far as I know, SPro has been successfully build and used on Linux, SPARC/SunOS, and HP-UX. It should also work on AIX though this has not been tested so far. This code has *never* been ported to any windows-based OS but if someone ever do that, please let me know (there should be some problems).


  To report a bug or any comment or (even better) any contribution, please send a mail to ggravier@irisa.fr.


  See CHANGES for details.


  Guillaume Gravier - ggravier@irisa.fr


  In alphabetical order: Raphael Blouet, Pierre Duhamel, Johnny Mariethoz, Sylvain Meigner, Alexey Ozerov, Jacques Prado.
Change Log
SPro 4.0 is *not* compatible with previous releases and exhibits major changes in its structure. The main difference is the introduction of feature streams (much alike the signal streams used for input). This new feature I/O mechanisms enables the processing of continuous (or very large) audio streams.

As introducing feature streams required a *major* redesign of the source code, all the library files have been profondly modified and, to avoid confusions, old spro_*.c names have been changed. Aslo, the SPRO FEATURE FILE FORMAT HAS CHANGED to accomodate for streams (e.g. number of frames unknown)...

It is impossible to describe all the changes bteween releases 3.3.1 and 4.0 but here are the main ones:

  - feature I/O via feature streams
  - redefined SPro feature file format (included the addition of an optional variable length ASCII header)
  - added LSF computation in slpc
  - added normalization qualifier (R)
  - delta, delta-delta, CMS can be computed directly in the main tools