Show & Tell Demonstration of ImmACS @ ICASSP 2016
ImmACS (Immersive Audio Communication System), FORTH’s spatial audio conference call system was presented at prestigious Show & Tell Session at ICASSP 2016, the largest international scientific conference on signal processing, in Shanghai. Specifically, the demonstration was presented in Session ST-1: Emerging Applications and Implementation Technologies, on Tuesday March 22, 2016. The participation of the audience was very positive; some photos of the demo are shown in this page.
ImmACS technology is largely based on the microphone technologies that FORTH has designed for the LISTEN project.
Technical information on the ImmACS demonstration follows.
ImmACS performs real-time 3D sound multi-participant teleconferencing. Our vision has been to design and implement a cost-effective telepresence platform, building from solid research foundations a system which is fun and natural to use. The audience members were able to communicate with each other using ImmACS in turns (wearing headphones); two clients were operating concurrently in various locations in the venue during the session.
The system performs the following tasks:
- Robust localization of multiple concurrent sound sources, performing direction-of-arrival (DOA) estimation using a circular microphone array.
- Sound source separation based on the estimated directions, enabling low-bitrate directional coding of the original soundfield into a downmix signal and side-information.
- Interactive reproduction using a GUI where the user can enhance/isolate sound(s) of interest, and/or move the sources anywhere around him/her.
- Multi-participant audio-visual communication, building on an existing open-source SIP user-agent. Through innovative solutions, our directional audio codec is transparent to the SIP platform; a server has also been developed supporting multiple clients.
- A microphone array is needed at each client. We demonstrated our custom-designed digital MEMS circular microphone arrays, promoting a cost-effective and compact solution (no soundcard needed).
For a video demonstration of the various system components see our YouTube channel: