Research
The RATEM project objective is to achieve both low-cost and high-quality 3D audio requirements to enable more realistic immersive teleprecencing and entertainment. To achieve the objectives, our development tasks and main research areas are listed below.
1. Design of biologically inspired miniature microphone array
- Design a low-cost four-sensor three-dimensional microphone array of minimum requirement of space separation biologically inspired;
- Implement sound capture interface to computer with pre-amp circuits, analogy-to-digital converter and the driver for sound receiver
- Calibrate the microphone array to retrieve the three-dimensional array response as function of frequency and direction of arrival
2. Real-time multiple acoustic sources 3D direction finding
- Investigate the methods for three-dimensional acoustic source direction finding using the new proposed miniature microphone array
- Implementation of the real-time 3D acoustic direction finding system including signal acquisition, voice activity detection, and direction estimation for 2D and 3D space searching and perform performance evaluation under various reverberation and noise levels
- Further investigate the effects of reverberation and noise on direction finding in meeting room and attempt to compensate for reverberation and noise for improved performance
- Study and develop reliable solution for direction finding of multiple acoustic sources, investigate methods of coherence test for spare time-frequency bins of the acoustic sources; investigate multiple source tracking approaches based on particle filtering/Kalman filtering
3. Real-time directional signal acquisition and enhancement
- Investigate the methods for reforming the array response to particular directions of interested acoustic sources
- Improve the output signal to noise and interference ratio using the proposed microphone array
- Implement the program of real-time capable and performance reliable
4. Real-time 3D audio reproduction
- Investigate the methods for reproducing the realistic 3D audio field using the proposed microphone array and appropriate playback system such as a stereo headphone or a group of directional louder speakers
- Investigate the effects of the head-related-transfer-function (HRTF) and room reverberations on the spatial hearing of humans
- Solve the well-known in-head and front/back confusion problems produced by the conventional approaches
Our audio research group is currently working together with the video and FPGA/GPU groups for integrating and implementing a realistic real-time audio-video telepresence.