Advanced low bit rate image and speech compression methods for portable communication terminals

• Forschungsstelle: COST
• Projektnummer: C97.0050
• Projekttitel Englisch: Advanced low bit rate image and speech compression methods for portable communication terminals

Schlüsselwörter (Englisch)

Wireless communications; image and low bit-rate video compression; low bit-rate speech coding; speaker recognition over GSM; multimedia processing; signal processing for very low-power VLSI implementation.

Forschungsprogramme (Englisch)

COST-Action 254 - Traitement intelligent et facilités pour les terminaux de communication

Kurzbeschreibung (Englisch)

See abstract

Weitere Hinweise und Angaben (Englisch)

Full name of research-institution/enterprise: Université de Neuchâtel Institut de Microtechnique Laboratoire d'Electronique et de Traitement du signal

Partner und Internationale Organisationen (Englisch)

B, CH, CRO, E, F, GR, H, I, IRL, N, NL, P, ROM, SK Sl

Abstract (Englisch)

The objective of the project is to provide effective algorithms and implementation architectures for future generations of wireless communication terminals, featuring (1) multimedia functions, such as integrated video, audio and data processing, (2) very low power consumption, in order to extend battery life and autonomy of terminals, and (3) miniature size, so that the devices can be held in one hand. Image and low bit-rate video compression: The image compression effort of this project was dedicated, during the reported period, to address two different topics. The first one consists in evaluating an alternati-ve method for image compression. This algorithm is based on implementations that take advantage of the rich amount of information that is present of the phase component of the image signals. The second effort was allocated to investigate computational complexity reduction strategies particularly useful for portable implementations. In this field, an architectural and an algorithmic approach have been studied. From the algorithmic point of view, co-operative work was made which consists in studying coding systems that use transformation kernels that require less arithmetical operations than the classical Discrete-Cosine- Transform-based methods. These schemes could be used in security/surveillance applications. Low bit-rate speech compression: The purpose is to study a low complexity variable bit rate (VBR) speech coder suitable for voice storage in portable devices, based on the CELP FS1016 standard with different algorithmic modifications aiming at reducing the computational complexity. A silence compression scheme was adapted to this coder, to achieve reduction in both the bit rate and the power consumption. Moreover, the new MELP coder (Mixed Excited Linear Prediction) FS1017 standard at 2.4 kbps was studied and implemented. It was found that although this coder has a similar quality than the FS1016, and a lower bit rate, it is unsuitable for the targeted low power application due to its high complexity. Speaker recognition and GSM speech coding: It is expected that in the near future many secured transac-tions (remote access to teleservices / telebanking) will take place through the mobile cellular phone. This part of the project aims at studying the influence of GSM coding on speaker recognition performance.

Datenbankreferenzen (Englisch)

Swiss Database: COST-DB of the State Secretariat for Education and Research Hallwylstrasse 4 CH-3003 Berne, Switzerland Tel. [phone] Swiss Project-Number: C97.0050