Short Description

There has been a long history of research investigating the effect of room acoustics, background noise and competing speakers on speech communication. On the one hand, research focused on estimating the cognitive load and the speech processing performance of listeners in specific noise conditions (e.g., on how many words can be recognized or memorized correctly), on the other hand, the individual response of a speaker that a specific vocal demand will cause. The impact the room has on communication is not always straight forward. Whereas a certain amount of reverberation supports the voice of the speaker, too much reverberation decreases speech intelligibility and increases noise. The aims of this project are 1) to analyze spontaneous dialogues taking place in an acoustically unfavorable room with background bubble talk with respect to acoustic speech characteristics that potentially reflect vocal fatigue and communication flow; 2) to improve the room acoustically (e.g., using absorption elements such as edge absorbers and/or poster panels) and 3) to analyze whether conversations taking place in the acoustically optimized room improve with respect to the acoustic speech measures.

Teams of 2-3 students are welcome!

Your Tasks (depending on specific project):

• Record speech data of approx. 5 participants
• Perform acoustic measurements in the unoptimized room and analyze it, using the appropriate acoustic measures
• Design an acoustic solution to improve the Room (specific requirement: in the room there will be lots of bubble talk, thus a class-room situation) and perform the acoustic measurements again
• Create a questionnaire to estimate the participants’ perception of the acoustic scene
• Adapt existing acoustic speech-feature extraction toolboxes available at SPSC Laboratory for the used materials
• Use Random Forest Classification to find those acoustic features that change in the speech before and after the room-acoustic improvement

Your Profile

• knowledge about room acoustics, room acoustic measurements and evaluation of room acoustic parameters
• good knowledge of programming (ideally Python)
• interest in the design and evaluation of listening tests

Contact:

Barbara Schuppler (b.schuppler@tugraz.at) Julian Koch (julian.koch@tugraz.at) Martin Hagmüller (hagmueller@tugraz.at)