Distance Distinguishing Microphone Array for Hands-free Interface

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

• Dr Yusuke Hioka (Department of Electrical and Computer Engineering, University of Canterbury, NZ)
• Wednesday 18 June 2014, 12:00-13:00
• Level 11: CAP Seminar Room.

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

Emphasising target speech is an important subject in hands-free audio-based interface utilised for, e.g. auditory sensors in robots and voice conference systems, because in general microphones also pick up the sounds made by various interfering sources around the speaker. In this talk, we focus on the use of microphone arrays and introduce two novel techniques that can separate sound whose sources are located in the same direction but at different distances from the microphone arrays.

The first method uses two small microphone arrays placed apart. The method is based on a novel algorithm, PSD (power spectral density) estimation in beamspace, which is able to estimate PSD of each sound source using diverse combination of the gain of multiple beamformers. With the pair of microphone arrays, the method estimates the PSD of sound sources located both inside and outside of a two-dimensional area. Then it emphasises the target speech by applying a post-filter, which is calculated from the estimated PSDs, to the received sound signal.

On the other hand, the second method uses only one small microphone array. The method distinguishes the distance of sources by using direct-to-reverberation energy ratio (DRR) as a clue since DRR has a one-to-one relation to the sound-source distances. A new model of the spatial correlation matrix used for estimating the DRR is introduced and the estimated DRR is utilised to emphasise the target speech.

In the last part of the talk a few developed products for voice conference, to which the research outcomes were implemented, are also introduced.

Speaker Biography

Yusuke Hioka has been a Lecturer at the Department of Electrical and Computer Engineering, University of Canterbury, Christchurch, New Zealand since 2013. Before he came to New Zealand, he worked at NTT Laboratories in Tokyo, Japan as a Research Engineer for nearly 8 years where he was mainly engaged in the research and development of audio / acoustic signal processing algorithm for voice conference systems.

Yusuke received his BE, ME, and PhD in Engineering from Keio University, Japan in 2000, 2002, and 2005 respectively. While he was at NTT , he was also a visiting researcher at the Victoria University of Wellington in 2011. His major research interests are audio/acoustic signal processing (particularly specialised in microphone arrays) and room acoustics. He is a senior member of the IEEE .

This talk is part of the COMMSP Seminar series.

This talk is included in these lists:

• COMMSP & CP list
• COMMSP Seminar
• Level 11: CAP Seminar Room
• SAP Talks

Note that ex-directory lists are not shown.