SOCRATES THEMATIC NETWORK IN PHONETICS AND SPEECH COMMUNICATION

SPOKEN LANGUAGE ENGINEERING WORKING GROUP

Survey of Courses in Spoken Language Engineering

Contributor(s):

Bernard GuÝrin

Institution:

Institut National Polytechnique de Grenoble (Ecole Nationale SupÝrieure d'Electronique et de RadioÝlectricitÝ de Grenoble)

Degree Course

Electrical Engineering.

Module title

TECHNOLOGY OF SPEECH

Level

Final year undergraduate or Diplîme d'Etudes Approfondies

Proportion of Student time

20 hours - Lectures - Time : October-December - ETCS Credits: 3

Aims and Objectives

Knowledge of speech signal properties. Explain attainable objectives in the field of speech analysis, speech coding and speech synthesis.

Prerequisites

* Signal processing theory (analog and digital signals). * Electrical circuit theory.

Syllabus

* Speech and communication.
* Theory of speech production.
* Methods of speech signal analysis.
* Linear prediction coding.
* Voice quality evaluation.
* Low bit-rate coding, medium bit-rate coding, high bit-rate coding.

Contact(s) at the institution

Bernard GuÝrin (guerin@icp.grenet.fr) ; Etienne Pic (pic@enserg.fr) Contributor(s): Bernard GuÝrin

Institution:

Institut National Polytechnique de Grenoble (Ecole Nationale SupÝrieure d'Electronique et de RadioÝlectricitÝ de Grenoble)

Degree Course

Electrical Engineering.

Module title

PATTERN RECOGNITION

Level

Final year undergraduate or Diplîme d'Etudes Approfondies

Proportion of Student time

14 hours - Lectures - Time : October-December - ETCS Credits: 1

Aims and Objectives

The goal is to provide the required knowledge to understand the two main approaches used in pattern recognition. At the end of the course, the student must be able to compare several possible solutions. He must be able to define the several steps which constitute the design of a pattern recognition system.

Prerequisites

* Probability theory.
* Statistical data analysis.
* Matrix calculus.

Syllabus

* Basic concepts and methods in mathematical pattern recognition.
* Decision function, cluster analysis and pattern classification by distance function.
* Automatic learning.
* Syntactic pattern recognition (concept from formal language theory, recognition grammar, grammatical inference.
* Speech recognition (Dynamic Time Warping, Hidden Markov Model).

Contact(s) at the institution

Bernard GuÝrin (guerin@icp.grenet.fr) ; Etienne Pic (pic@enserg.fr) Contributor(s): Gang Feng

Institution:

Institut National Polytechnique de Grenoble (Ecole Nationale SupÝrieure d'Electronique et de RadioÝlectricitÝ de Grenoble)

Degree Course

Electrical Engineering.

Module title

SPEECH AND AUDIO CODING

Level

Final year undergraduate or Diplîme d'Etudes Approfondies

Proportion of Student time

20 hours - Lectures - Time : October-December - ETCS Credits: 3

Aims and Objectives

This course is intended for engineering students involved in the design and the application of digital speech and audio coding systems ; it provides comprehensive understanding of both the fundamental theories and the different techniques used in many applications of modern communication systems.

Prerequisites

* Basic concepts of digital signal processing.
* Basic concepts of information theory.

Syllabus

* Redundancy in speech and audio signals.
* Uniform, nonuniform, optimal and adaptive quantisation.
* Predictive coding.
* Sub-band coding and transform coding.
* Noise shaping with psychoacoustic models.
* Objective and subjective quality measures.