/speech_tools

Edinburgh Speech Tools

Primary LanguageC++OtherNOASSERTION

speech_tools EDINBURGH SPEECH TOOLS LIBRARY VERSION 2.5.1 July 2020

Principal authors: Alan W Black, Paul Taylor, Richard Caley, Simon King, Rob Clark and Korin Richmond

Centre for Speech Technology University of Edinburgh Informatics Forum 10 Crichton Street Edinburgh EH8 9AB, UK http://www.cstr.ed.ac.uk/projects/speech_tools https://github.com/festvox/speech_tools/

Although this is a considered stable release, there are still a number of known parts are missing or incomplete.


The Edinburgh speech tools system is a library of C++ classes, functions and utility programs that are frequently used in speech software. The system compiles to a single unix library .a file which can be linked with software. At present, C++ classes for several useful speech and language classes have been written, along with audio software and some basic signal processing software.

========================================================================

        Edinburgh Speech Tools Library version 2.5.1
          Centre for Speech Technology Research                
               University of Edinburgh, UK                     
                Copyright (c) 1994-2020
                  All Rights Reserved.                         

Permission is hereby granted, free of charge, to use and distribute
this software and its documentation without restriction, including
without limitation the rights to use, copy, modify, merge, publish,
distribute, sublicense, and/or sell copies of this work, and to
permit persons to whom this work is furnished to do so, subject to
the following conditions:

  1. The code must retain the above copyright notice, this list of
    conditions and the following disclaimer.
  2. Any modifications must be clearly marked as such.
  3. Original authors' names are not deleted.
  4. The authors' names are not used to endorse or promote products
    derived from this software without specific prior written
    permission.

THE UNIVERSITY OF EDINBURGH AND THE CONTRIBUTORS TO THIS WORK
DISCLAIM ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING
ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT
SHALL THE UNIVERSITY OF EDINBURGH NOR THE CONTRIBUTORS BE LIABLE
FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN
AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION,
ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF
THIS SOFTWARE.

========================================================================

The following c++ programs are available:

na_play: generic playback program for use with net_audio and CSTR ao.

ch_wave: Waveform file conversion program.

ch_lab: label file conversion program.

ch_track: Track file conversion program.

    wagon: a CART tree build and test program

    And others

The following C++ sub-libraries are available

audio:	C++ audio functions for Network Audio system, Suns
            Linux and FreeBSD

speech_class:	C++ speech classes, including waveform and track.

ling_class:	C++ linguistic classes.

sigpr:	Signal processing

utils: Various utilities.	

COPYING

Some files in the distribution do not fall under the above copyright. However none have any commercial restrictions, and no files fall under the GPL. Specifically files that have different licences are as follows:

grammar/wfst/wfst_train.cc Copyright Carnegie Mellon University (under the same open licences as much of the rest of the code). siod/ include/siod*.h lib/siod/siod.scm Apart from some new functionality under our licence the SIOD files fall under the licence given in siod/slib.cc copyright Paradigm Associates. This code has been substantially modified by us though our changes are under that licence.

    editline.c editline.h el_complete.c el_sysunix.c el_unix.h
    are copyright 1992 Simmule Turner and Rich Salz under the
    "BSD like" licence in siod/editline.c  Again substantial 
    changes have been made by us which continue to be under that
    licence

rxp/
include/rxp/ Is Copyright Richard Tobin. We have a specific version of RXP which allows free (including non-commercial) use.

base_class/raveconv.cc Is copyright 1992, 1995 by Markus Mummert

base_class/string/reg*.{c,cc} Copyright (c) 1986 by University of Toronto.

base_class/string/EST_strcasecmp.c Copyright (c) 1987, 1993 The Regents of the University of California. All rights reserved.

The standard BSD licence

ACKNOWLEDGEMENTS

Most of the software was written by Paul Taylor, Alan W Black, Simon King, Richard Caley, Rob Clark and Korin Richmond. Additional contributions have been made by the following people. Slightly different licence agreements accompany these pieces of code, so it is important that the various agreements be properly understood.

Tony Robinson: some signal processing code. Markus Mummert: sample rate converter (rateconv.cc). Craig Reese and Joe Campbell: ulaw conversion code. Paul Bagshaw: pitch tracker. Theo Veenker: IRIX audio support. Stan Chen: AIX support Paradigm Assoc. and George Carrett: For Scheme In One Defun Simmule Turner and Rich Salz: for editline Henry Spencer: for regex code The Regents of the University of California: some string comparison code Richard Tobin: RXP, XML parser. Arthur Toth: Windows server code Pierre Lorenzon devel@pollock-nageoire.net gcc-4.7 patches Sergio Oller: for making the documentation work again

INSTALLATION


see INSTALL file for details

FILE/DIRECTORY STRUCTURE

The main directory contains minimal wrap-around main() functions which compile to form executables which use the algorithms in the speech tools. All these main() functions really do is to read command line options and read and write files, the algorithms themselves are completely contained within the libraries.

The directory structure is as follows:

config: shared make rules and configuration make rules

speech_class: general speech class files, including waveforms, contours etc.

ling_class: general linguistic class files, including a stream-based architecture.

sigpr: signal processing code.

doc: basic documentation, generates html, info and postscript

include: class and general header files.

lib: where the .a files are kept

main: Directory containing main files which compile to executables.

DOCUMENTATION

There are two sources of documentation.

Man pages exist for the more important stand-alone programs. These are found in the main directory.

The library documentation has substantiall improved in this version using an integrated DOC++, JADE and docbook combination. Classes are described at the overview, theorectical, practical and code level. The build process for the documentation is still dependent on some more obscure packages so it is recommended that you take the pre-built documentation. Or access it on-line through http://www.cstr.ed.ac.uk/projects/speech_tools

The library documentation is in texinfo format which can be converted into GNU info HTML and postscript, see Makefile in doc/

Note the documentation is not up to date. We are in the process of moving the documentation to Doxygen.

BUGS/COMMENTS ETC

Substantial parts of this release are considered mature and should be relatively free of bugs, however there are newer parts which have not been rigourously tested yet. If you find any bugs, please send a bug report to speechtools@cstr.ed.ac.uk. Any useful comments will also be appreciated. See also the bugs file in the doc directory.

If you have some software that you think could be integrated into the speechtools system, please contact us at speechtools@cstr.ed.ac.uk. The whole point of writing and releasing code such as this is to prevent duplication of effort, and therefore we would appreciate greatly any code which will add to the usefulness of the system. Signal processing routines are particularly welcome.

ADDITIONAL SOFTWARE

CSTR supplies a number of different speech technology systems including speech synthesis systems and (in the future) a recognition system. To find out more information checkout http://www.cstr.ed.ac.uk/