Logo multitel

ASR-Voice recognition

Automatic speech recognition is used in many applications.
These applications make use of speech in order to replace or complement an interface to communicate with a machine, e.g. to access a service or control a functionality of an equipment. Indeed in some cases when the speaker cannot touch the equipment, speech is the only possible interface (eg: hands-free applications).

Multitel activities in Automatic Speech Recognition (ASR) are focused on applied research, software development and consulting.

Applied Research

This activity consists in studying mathematical theories, algorithmic concepts and computer tools that are required for developing ASR software. Multitel's current research in automatic speech recognition includes the following topics: robust acoustical analyses (e.g. robustness to background noise and room reverberation), technology for automatic phonetic transcription, dialogue strategy optimisation, as well as Distributed Speech Recognition (DSR).

Software Development

Research at Multitel in Automatic Speech Recognition has led to the development of the STRUT (Speech Training and Recognition Unified Tool) software.
STRUT is based on the OOBP (Object Oriented Block Programming) paradigm and provides a framework to develop and test ASR algorithms.
The corporate company Acapela is marketing software products for Automatic Speech Recognition that are based on STRUT.


Besides research and development activities, Multitel provides consulting in Automatic Speech Recognition : technical training for engineers and software developers, ASR application design (specifications and prototyping), speech data recording to train and test ASR systems, support of ASR systems under operation.