Speech intelligibility enhancement in large social rooms by means of computer modelling

OData support
Supervisor:
Dr. Márki Ferenc
Department of Networked Systems and Services

In spite of the many new research and improvements in both the field of acoustics and sound reinforcement system design one can find many large public spaces, with even newly designed sound reinforcement systems, where it’s difficult to understand speech. In this present study the objective approach of room acoustics is used to conclude which measures can be taken to reach proper speech intelligibility in practice.

The main objective of this study is to provide solutions for a chosen large room with poor intelligibility, and to prove the effectiveness of the resulting design both theoretically and with the aid of computer modelling and simulations.

To reach this, besides providing the necessary sound pressure level, target values for reverberation time below 1.5s, C50 clarity values of 0dB, and an STI speech transmission index value of 0.54 as recommended for churches, but at least a value of 0.45 to reach a generally fair intelligibility category were set, in accordance with international standards ISO 3382 and IEC 60268.

By means of ray tracing simulations it can be found out how much improvement is needed for each of the above parameters. After describing the ideas for solution based on theory, simulations for specifically these cases were run to conclude how much each of these changes contribute to speech intelligibility, with regard to the just noticeable difference of the investigated parameters.

The results and conclusions in case of the particular room studied could be summarized as follows: The maximum reachable sound pressure level before feedback can best be increased by decreasing the distance of the speaker to the microphone, target values for the other parameters can be reached by mainly applying sound absorbing material and using highly directive column speakers set to cover only the audience areas. Muting unused microphones, using microphones and loudspeakers with a more optimal frequency response and isolation from outside noise sources and reverberant coupled rooms lead to further improvement of intelligibility.

In the end, it can be concluded, that speech intelligibility in the investigated room can be sufficiently improved by means of the discussed measures, and that this improvement can be shown with the use of ray tracing simulations.

Downloads

Please sign in to download the files of this thesis.