Accueil > Recherche > Démonstrations
Comparison of French TtS with the same sentence
Examples of French-speaking TtS with comparable sentences. Prestored files presented below are in a SUN/NEXT format, at Fs=8kHz, MuLaw coding.
| "Je suis le synthétiseur
de l'Institut de la Communication Parlée. Je convertis un texte
libre en signal acoustique. Merci de votre attention. " |
|
 |
ICP-Grenoble (F), 1993 |
| "Je suis le synthétiseur
du CNET. Je convertis un texte libre en signal acoustique. Merci de votre
attention." |
|
|
CNET-Lannion (F), 1993 (TD-PSOLA) |
| "Je suis le synthétiseur
du KTH. Je convertis un texte libre en signal acoustique. Merci de votre
attention." |
|
|
KTH-Stockholm (S), 1993 |
| "Je suis le synthétiseur
du LAIP, et je convertis un texte libre en signal acoustique. Je vous
remercie de votre attention." |
|
|
LAIP-Lausanne (CH), 1996 (MBR PSO) |
| "Je suis un système
de synthèse à partir du texte. J'ai été conçu
dans les laboratoires de la Faculté Polytechnique de Mons" |
|
|
University-Mons (B), 1993 (LPC) |
|
University-Mons (B), 1993 (MBE) |
|
University-Mons (B), 1993 (MBR PSO) |
|
University-Mons (B), 1993 (TD PSO) |
Phoneme durations have been obtained from a human reading, from which pitch
has been measured and then grossly stylized.
LPC = Linear Predictive Coding
MBE = Multi-Band Excited (frequency band analysis - all signals in the bands
are phase sync.)
MBR = Multi-Band Re-synthesis (diphones are stored with constant pitch in the
dictionary. Then Fo is controlled by a TD-PSOLA
Audio-Visual Speech: examples of TtS-generated talking heads
Examples of Text-to-AUDIO-VISUAL-Speech Synthesis The movies presented below are in a SGI or QuickTime format, at 25 images per sec.
| "Comme vous le voyez,
je sais lire automatiquement un texte, mais je n'ai pas de visage ; je
n'en ai pas besoin pour etre compris" |
|
 |
SGI format (3 MB) |
|
QuickTime format (1.7 MB) |
| "Je suis le synthétiseur
audiovisuel de l'Institut de la Communication Parlée." |
|
|
SGI format (1.6 MB) |
|
QuickTime format (1.3 MB) |
| Strange example TtAVS generating
pseudo-Japanese from a Japanese-like input written in French! by Bertrand
Le Goff, |
|
|
SGI format (3 MB) |
|
GZ compressed QuickTime format (2.8 MB) |
| "Je suis animé
par analyse video d'un locuteur qui prononce le texte que vous entendez." |
|
|
SGI format (4 MB) |
Links to on-line interactive systems
The only way to evaluate by yourself the actual quality of a system: No cheating
here!!!
– Multilingual TtS (Swedish, English, French, German, Spanish)from KTH,
S
– Multilingual TtS (English, German, Chinese)from Bell Labs / Lucent Tech.,
USA
– Multilingual TtS (German, English, Japanese)from Univ. Duisburg, D
– TtS in French from ICP-Grenoble, F
– "Hadifix" TtS in German from IKP-Bonn, D
– TtS in English from AT&T Labs, USA
– TtS in English from Univ. Twente, NL
– TtS in English from Entropics, US
– TtS in GermanSVOX, from TIK, CH
– TtS in Englishfrom CSTR, UK
– TtS in EnglishTruvoice by Centigram, USA
– "Laureate" TtS in English from British Telecom, US
– TtS in US EnglishModel Talker, from ASEL, Univ. Delaware, USA
Links to examples of Pre-stored audiofiles
– The YORKTALK system in English from the University of York
– The KTH multilingual system from the KTH-Stockholm
– EUROVOCS, Gent, Belgium
– Speech Museum at Univ. of Birmingham
– Examples of TtS from HMMs at Cambridge Engineering, UK
– TTS in French, English, German & Spanish+ free download for the PCs
by ELAN Informatique, France
– The MBROLA projectinitiated by the University of Mons, B
– TTS in English by Microsoft Research, USA
– MacinTalk TTS in English & Spanish+ free download for the Macs by
Apple, USA
– TTS in English by SoftVoice Inc., USA
– TTS in German by Univ. Saarlandes, D
– TTS in US E nglish by Eloquent Tecnology, USA