The MBROLA Project
Towards a Freely Available Multilingual Speech Synthesizer

1996 - 2006: 10 years! Happy Birthday MBROLA
MBROLA reads EmbROlA/, think of "umbrella"...

Winner of the IT European Award 1996
The aim of the MBROLA project, initiated by the TCTS Lab of the Faculté Polytechnique de Mons (Belgium), is to obtain a set of speech synthesizers for as many languages as possible, and provide them free for non-commercial applications. The ultimate goal is to boost academic research on speech synthesis, and particularly on prosody generation, known as one of the biggest challenges taken up by Text-To-Speech synthesizers for the years to come.
Central to the MBROLA project is MBROLA, 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 used (it is therefore NOT a Text-To-Speech (TTS)synthesizer, since it does not accept raw text as input). This synthesizer is provided for free, for non commercial, non military applications only.
Diphone databases tailored to the Mbrola format are needed to run the synthesizer. A French voices have been made available by the authors of MBROLA, and the MBROLA project has itself been organized so as to incite other research labs or companies to share their diphone databases. The terms of this sharing policy can be summarized as follows :
After some official agreement between the author of MBROLA and the owner of a diphone database, the database is processed by the author and adapted to the Mbrola format, for free. The resulting Mbrola diphone database is made available for non-commercial, non-military use as part of the MBROLA project. Commercial rights on the Mbrola database remain with the database provider for exclusive use with the Mbrola software.
MBROLA synthesis demos
The Mbrola Christmas song (in mp3 format), fully synthestic choral music!
Melissa , an MBROLA-based Virtual Singer
- Afrikaans (af1 male voice, 181K Wav file)
- American English (us1 female voice, 198K Wav file)
- American English (us2 male voice, 156K Wav file)
- American English (us3 male voice, 156K Wav file)
- Arabic (ar1 male voice, 297K Wav file)
- Arabic (ar2 male voice, 229K Wav file)
- Brazilian Portuguese (br1 male voice, 200K Wav file
- Brazilian Portuguese (br4 female voice, 260K Wav file))
- Breton (bz1 female voice, 137K Wav file)
- British English (en1 male voice, 340K Wav file)
- Canadian French (ca1 male voice, 187K Wav file)
- Canadian French (ca2 male voice, 616K Wav file)
- Croatian (cr1 male voice, 87K Wav file)
- Czech (cz1 female voice, 160K Wav file)
- Czech (cz2 male voice, 150K Wav file)
- Dutch (nl2 male voice, 115K Wav file)
- Dutch (nl3 female voice, 127K Wav file)
- Estonian (ee1 male voice, 183K Wav file)
- Spanish(es1 male voice)
- Spanish(es2 male voice)
- Spanish(es3 female voice)
- Spanish(es4 male voice)
- French(fr1 male voice, 156K Wav file)
- French(fr2 female voice, 143K Wav file)
- French(fr3 male voice, 180K Wav file)
- French(fr4 female voice, 75K Wav file)
- French(fr5 male voice, 75K Wav file)
- French(fr6 male voice, 156K Wav file)
- French(fr7 male voice, 156K Wav file)
- German (de1 female voice, 133K Wav file)
- German (de2 male voice, 169K Wav file)
- German (de3 female voice, 203K Wav file)
- German (de4 male voice, 360 K Wav file)
- German (de5 female voice, 413 K Wav file)
- German (de6 male voice, 204 K Wav file)
- German (de7 female voice, 298 K Wav file)
- German (de8 male voice, 143K Wav file)
- Greek (gr1 male voice, 54K Wav file)
- Greek (gr2 male voice, 250K Wav file)
- Korean (hn1 (hanmal) male voice, 109K Wav file)
- Hebrew (hb1 male voice, 287K Wav file)
- Hebrew (hb2 female voice, 260K Wav file)
- Hindi (in1 male voice, 180K Wav file)
- Hindi (in2 female voice, 207K Wav file)
- Hungarian (hu1 female voice, 257K Wav file)
- Icelandic (ic1 male voice, 154K Wav file)
- Indonesian (id1 male voice, 130K Wav file)
- Iranian (ir1 male voice, 153K Wav file)
- Italian (it1 male voice, 145K Wav file)
- Italian (it2 female voice, 175K Wav file)
- Italian (it3 male voice, 172K Wav file)
- Italian (it4 female voice, 193K Wav file)
- Japanese (jp1 male voice, 287K Wav file)
- Japanese (jp2 female voice, 104K Wav file)
- Japanese (jp3 female voice, 200K Wav file)
- Classical Latin (la1 male voice, 188K Wav file)
- Lithuanian (lt1 male voice, 106K Wav file)
- Lithuanian (lt2 male voice, 105K Wav file)
- Malay (ma1 female voice, 65K Wav file)
- Polish (pl1 female voice, 247K Wav file)
- Portuguese (European) (pt1 female voice, 168K Wav file)
- Romanian (ro1 male voice, 168K Wav file)
- Spanish Mexican (mx1 male voice, 73K Wav file)
- Spanish Mexican (mx2 male voice, 73K Wav file)
- Swedish (sw1 male voice, 100K Wav file)
- Swedish (sw2 female voice, 160K Wav file)
- Telugu (tl1 female voice, 87K Wav file)
- Turkish (tr1 male voice, 142K Wav file)
- Turkish (tr2 female voice, 142K Wav file)
- Spanish Venezuelan (vz1 male voice, 117K Wav file)
Comparison with other synthesis methods.
Introduction to TTS synthesis
For those of you who wish to learn more about Text To Speech technologies,
check this short introduction to TTS synthesis or, for more details get T. Dutoit's book.

Last updated January 4th, 2005, send comments to thierry dot dutoit "at" umons dot ac dot be
|