Géza Gordos

A new degree of freedom in ATM network dimensioning: optimizing the logical configuration

A mathematical model is presented that provides a well-defined formulation of the logical configuration problem of ATM networks (the carriers of future B-ISDN) with the objective of maximizing the total expected network revenue, given the physical network parameters and the traffic requirements of each virtual subnetwork. A two-phase solution procedure is developed in which the decision variables are the logical link capacities that specify the logical decomposition into virtual subnetworks, and the load sharing parameters. The first phase of the solution finds a global optimum in a rougher model. The second phase uses this as an initial point for a gradient-based hill climbing that applies the partial derivatives of the network revenue function obtained in a more refined model. >

Profivox—A Hungarian Text-to-Speech System for Telecommunications Applications

The latest Hungarian text-to-speech (TTS) system developed for telephone-based applications is described. The main features are intelligible human-like voice; robust software designed for continuous running; fully automatic conversion of declarative (short and very long) sentences and questions; and real time parallel operation, running on minimum 30 channels. The concept of prosody generation and sound duration processing is introduced. Also, the development environment of Profivox is presented. The market-leader Hungarian mobile service provider applies the TTS system in an automatic e-mail reading application.

Language processing for name and address reading in Hungarian

Name and address reading is an important combined application area of language processing and text-to-speech (TTS) systems. It is the cornerstone of both traditional reverse directory telephone services and new, location based, traffic and tour guide applications. The language processing aspects of a solution for Hungarian is described. The work was based on the analysis of a subscriber database containing about 3 million records (there are about 10 million Hungarian citizens). Categories of name and address elements were defined. A program for the automatic classification of database records was developed. Statistical parameters were derived about proper/legal names and addresses. Based on these results text corpora for enriching the TTS acoustic database were designed. Reading strategies and related special algorithms and tables were developed for the description of complex name categories. Our results may be applied for similar tasks of other languages with comparable linguistic and statistical features.

Objective Speech Quality Estimation for Analog Mobile Channels: Problems and Solutions

This article describes an objective speech quality assessment method developed for the Hungarian NMT-450 cellular mobile telephone system. The method is based on a psychoacoustic front end followed by a cognitive modeling component. Special problems of the NMT system, such as handover (HOV), the effects of automatic gain control (AGC), and the intrusion of signaling noise, are addressed in the cognitive module. Correlation between the subjective and objective quality measures is maximized by finding a transformation that linearizes their relationship. A correlation of 0.95 is achieved on an independent test set between the subjective speech quality and the proposed objective quality measure.

Scatternet Formation in High-Rate Wireless Personal Area Networks by Integer Linear Programming

Scatternet formation is one of the most important issues of emerging wireless personal area networks (WPANs). Numerous papers pointed out that the performance of a WPAN highly depends on the scatternet structure; however, for next generation WPANs (IEEE 802.15.3 and 802.15.4) the issue of scatternet formation is not discussed in the literature so far. This paper presents the techniques, constraints and effects of scatternet formation. First, the theoretically based problem is formulated in integer linear programming (ILP) framework. Two heuristic methods are proposed for high-rate WPANs (802.15.3), which can be extended for low-rate WPANs (802.15.4) as well. The constructed networks are analyzed based on two metrics: (1) total network load, (2) amount of piconets. We show that network performance highly depends on the used algorithm: the network load can be reduced by up to 30% by using a proper scatternet formation algorithm. Furthermore, the scatternet can be optimized depending on the actual network traffic and individual objectives

Design of a Hungarian Emotional Database for Speech Analysis and Synthesis

This paper reports on the ongoing work of designing and building a Hungarian emotional speech database. First, a brief overview of the related earlier work on the subject is given. Next, the targeted applications of our emotional speech database are described. The problem of creating or collecting suitable prompts for different emotions and speaking styles is addressed. Then, we discuss the problem of collecting material for child tale reading. Finally, we present the methods of database validation and annotation.