This paper describes Interactive Speech Self-Service System (ISSS) that provides a framework for building voice applications. An example of voice application is shown in this paper. The application enables telephone subscribers to obtain a flight schedule and flight status information from the remote public web-service of Aeroflot company. The possibility of using remote sources of information for interactive telephone dialogs is shown. The system is built using state-of-the-art technologies such as SIP for telephone infrastructure, VXML for dialog control, ASP.NET for dynamic content generation, web-services SOAP for information access, speaker-independent speech recognition technology for understanding user input. In this paper we also present the review of dialogue optimization methods, which means improving success rate and reducing number of turns in dialogue. To understand this methods we also included speech dialogue classification and emphasized differences between direct dialogues and natural speech understanding. The distinctive features of Russian language are also concerned. Further development of ISSS system will include development of voice applications using natural speech recognition techniques.