Project STEEM (Speech Transmission End-to-End Monitoring)

The project was supported by the central innovation programme for small and medium-sized enterprises (Zentralen Innovationsprogramm Mittelstand, ZIM).

Project duration: Oct. 2014 - May 2018

Project description

The goal of the project is to develop a parametric model which will be able to assess the speech transmission quality of VoIP. The important feature of the parametric model is that it does not have access to the speech data itself but only to the parameters of the network and end-point VoIP terminals. This is important as it does not raise any privacy concerns and is easier to apply in telecommunication networks.

Under the project a range of subjective tests are designed and conducted where the test participants are asked about quality of different simulated degradations typical to Voice over IP (VoIP) technology (e.g. speech coding, packet loss, delay and echo). The main tests will be simulated conversations which are resource and time consuming but reflect the real use case of VoIP. What cannot be covered by these tests will be addressed by much cheaper listening tests where many more conditions can be evaluated in shorter time. The results of all experiments will be merged and analyzed to derive a quality assessment model.

Project partners

• HEAD acoustics GmbH (Herziogenrath, www.head-acoustics.de/de/index.htm)
• AVM computersysteme Vertriebs GmbH (Berlin)
• Technische Universität Ilmenau, Institute for Media Technology, Audiovisual Technology Group

Project leader:  Prof. Dr.-Ing. Alexander Raake

Project staff: M.Sc. Michał Sołoducha, Dr.-Ing. Janto Skowronek, Dr. Eckhardt Schön