Learning the Reverberation Domain

Add to your list(s) Download to your calendar using vCal

• Constantinos Papayiannis (PhD student, speech and audio processing)
• Monday 04 September 2017, 12:00-13:00
• TBD.

If you have a question about this talk, please contact Alastair Moore.

Reverberation is present in all real life enclosures. From our workplaces to our homes and even in places designed as auditoria, such as concert halls and theaters. We have learned not only to understand speech and sounds in the presence of reverberation but also to use it for aesthetics in music and adjust it to enhance our understanding of speech.

This understanding of speech and sounds in the presence of reverberation is not inherited by machines designed for understanding us. Automatic Speech Recognition (ASR) and other machine-human interaction systems suffer a decrease in performance in the presence of reverberation. This problem inspired this work. The aim is to find solutions to this problem by creating frameworks which enable machines to learn reverberation through examples. These frameworks would allow machines to better understand speech in reverberant environments and also to understand reverberation itself. The ability to infer properties of the enclosure is one aspect of this, which can involve inferring the size of the room and even its type.

This talk is part of the COMMSP Seminar series.

This talk is included in these lists:

• COMMSP & CP list
• COMMSP Seminar
• SAP Talks
• TBD

Note that ex-directory lists are not shown.