Bug#509517: ITP: mbrola -- multilingual software speech synthesizer
Version: N/A; reported 2008-12-21
* Package name : mbrola
Version : 3.01h
Upstream Author : Dr Thierry Dutoit <email@example.com>
* URL : http://tcts.fpms.ac.be/synthesis
* License : see the file readme.txt in the source zip: non-free as in
without source code, and for non-commercial, non-military
purposes, with and only with the voice and language
databases made available by the author.
Description : multilingual software speech synthesizer
Mbrola is Thierry Dutoit's phonemizer for multilingual speech synthesis. The
various diphone databases are distributed on separate packages, but they
must be used with and only with Mbrola because of license matters. Read the
copyright for details.
Mbrola itself doesn't provide full TTS. It is a speech synthesizer based on
the concatenation of diphones. It takes a list of phonemes as input,
together with prosodic information (duration of phonemes and a piecewise linear
description of pitch), and produces speech samples on 16 bits (linear),
at the sampling frequency of the diphone database.
Use Mbrola along with Freephone, cicero or espeak to have a complete
text-to-speech in English.
I'm actually taking oralux' packaging of mbrola to Debian.
Yes, that would basically be a fully binary package (i386, amd64, ppc, arm,
alpha, and sparc), no source, no way to modify it (even the voices), and I had
to get permission from the author to distribute it. However, the quality of the
voices is very good compared to what we currently have in the main section.
The idea would be to have the mbrola package containing the binary, and then
separately package voices as mbrola-xyn, containing voice xyn (i.e. mbrola-fr1,
mbrola-fr2, mbrola-en1, mbrola-en2 etc.) Since voices are a few MB, that is not
really a waste to have separate packages, and permits lighter updates.
I intend to also package freephone, cicero etc. but won't do so if mbrola itself
can not enter even non-free.