Main objectives: the “Speech Communication” Department mostly performs fundamental research, while also keeping an eye on applicative development. The scientific interest of the department concentrates into studies of linguistics, phonetics and phonology of Vietnamese, and other languages in Vietnam and in the near region, in order to better understand the mechanisms of production/perception/understanding, and in order to transfer these fundamental results to the development of high technology in Vietnam. For several years, the department addresses simultaneously fundamental research and more applied research through Natural Language Processing (NLP) and Vocal Technologies. However, the department researchers aim not to study all possible topics in NLP and speech processing (a very large research field), but they aim to provide the MICA Institute with indispensable tools in order to realize, through the use of multimedia, speech communication tools in Vietnamese, Khmer and Lao, useful for man-machine or computer-aided man-man dialogue.
Indeed, these tools will be very useful for Vietnamese industrials that wish to modernize and integrate speech technology in their tools or application

 

MICA Speech Communication Department is the only Vietnamese research team leading simultaneously both research activities: natural language processing and vocal technologies.

 

Research axes and activities:
If it seems natural that “Speech Com” researchers study the Vietnamese standard language (ethnic Viet/Kinh), it is nevertheless important that they are also interested in other languages spoken in Vietnam. Indeed, these languages present common origins (or at least regional characteristics) and are mostly tonal languages. From a scientific point of view it is fundamental to understand the specific involved mechanisms and attempt to extract generic models.

However, it is vital for Vietnam to have technological tools and applications using the Vietnamese language (and other languages of the region). These tools are strongly involved in the development of information technologies in Vietnam and the region. They already play a very important role for socio-economic development. These technologies, known as vocal technologies, are often necessary in the development of human-machine interaction systems.

Another scientific motivation is the development of its research activities under the “multilingual technology” field, developing multilingual machine translation tools or multilingual synthesis/recognition tools (with the Vietnamese as one of the languages covered).

 

 

 

 

MISTRAL+ (Melody Intonation Speaker Tonal Range Analysis using variable Levels) is a dedicated tool for the analysis of the expressive speech of any language (accented, stressed, tonal or not) at the lexical/intonation levels, and also the specific melody of tonal languages (acoustic level). MISTRAL+ is also designed for the study of under-resourced languages.

HoaSen & Hoa Sung are respectively MICA ASR toolkit and synthesis engine, both designed for different platforms (PC, embedded/Android, DSP), used for H-M interactions.

MISTRAL+ snapshot
       

MO PIU recording session
Mo Piu language
In the framework of AU CO project, MICA researchers started the study of one of minority language, spoken by 237 people only, undocumented and without a writing system. This endangered language, from the North Vietnam, seems a branch still unknown of the Hmong-Mien family. Several recording session were conducted, thereby providing a first database composed of 36h films, 35h speech, 1h songs, for a total of 4000 sound/video files.

 

 

Projects in process
- RECO²CONTROL (2nd phase)
- ANR Blanc APPSy (Asymétries Phonétique et Phonologique de la Syllabe)
- ÂU CƠ (AUdio COrpus of Vietnamese minority languages)
- APRIL (Avancées dans le Prototypage Rapide pour l’Informatisation des Langues)
- KRAP (Khmer Rouge Archives Preservation)

Completed projects
- TALK (Traitement Automatique de la Langue Khmère),
- CORUS
- VLSP (Vietnamese Language and Speech Processing)
- Smart Robot
- ANR Blanc PI Languages

Network leading/participating
- CONE (Collaborative Open Network for Education),
- SPAN (Speech Processing Asian Network)
- COLIPS - ISCA - AFCP

Contacts
Trần Đỗ Đạt (Director) – This email address is being protected from spambots. You need JavaScript enabled to view it.
Nguyễn Việt Sơn (Vice-director) - This email address is being protected from spambots. You need JavaScript enabled to view it.
Alexis Michaud – This email address is being protected from spambots. You need JavaScript enabled to view it.
Eric Castelli – This email address is being protected from spambots. You need JavaScript enabled to view it.
Mac Dang Khoa – This email address is being protected from spambots. You need JavaScript enabled to view it.
Do Thi Ngoc Diep – This email address is being protected from spambots. You need JavaScript enabled to view it.
Nguyen Cong Phuong - This email address is being protected from spambots. You need JavaScript enabled to view it.

 

Partners
IOIT (Vietnam)
SoICT (Vietnam)
LIG (France)
GIPSA-Lab (France)
Telecom Paris (Fr)
LPL (France)
PRAXILING (France)
DDL (France)
IA (France)
LORIA (France)
LPP (France)
LIMSI (France)
ORANGE (Fr,Jp)
NII (Japan)
ITC (Cambodia)
USM (Malaisia)