EINDHOVEN, The Netherlands. Philips Research is exploring ways to improve the audio quality of consumer video conferencing services done over the Internet using a TV set. The work is one small part of efforts smoothing the way to bring the TV into the Web era.
At the Consumer Electronics Show in January, several top TV makers showed prototypes of their latest sets using Skype for video conferencing. The service was one aspect of an emerging wave of Web-connected TVs.
"We think there are a few issues to tackle to have good quality communications," said Kees Janse, a principal scientist in digital signal processing at Philips Research. "Consumers will tend to have longer, more relaxed conversations when using a TV so we think it's important to solve these problems," he added.
Specifically, the group has developed code to eliminate reverberation affects, localize audio in 3-D space and selectively filter speakers. It is also working on ways to use its stereo 3-D video technology to enhance conferencing.
The reverb problem comes from the fact users will typically sit four meters from a TV and talk in a relatively low voice. In some rooms, audio reflections may swamp the sound of the voice, causing listening fatigue in as little as 15 minutes.
Philips researchers tackled this and other audio issues by using an array of microphones that estimate audio transmission paths. Adaptive beam forming algorithms on DSPs filter signals, focusing on the direct channel. Janse demos the Philips software in the video below.
Researchers used similar techniques to isolate individual speakers. Once captured, the audio streams can be placed in 3-D audio space. In addition, a secondary speaker can be filtered out when he talks over a primary speaker.
"We want this to work in people's living rooms, and we know we can't put any limits on the rooms these products go into," said Janse. "So we need a lot of adaptive beam forming and echo canceling in our digital signal processing," he said.
The team has developed software on PCs to handle the audio issues and is now porting that to select audio DSPs used in its TV sets.
"The software is really robust, and works in all acoustic situations," said Janse. "We think the problems are solved in such a way that mass adoption is possible and will come quite quickly," he said.
One wrinkle is that no common applications program interface exists for echo cancellation devices. Philips has worked with Skype to be able to turn off the Skype echo cancelation so it does not conflict with the TV maker's own approach.
Philips is also working on ways to use its lenticular lens display technology to provide stereo 3-D effects in video conferencing. Specifically, they want to create an effect that one speaker is looking or pointing directly at another and not at a third person.
Janse and colleagues at Philips show the audio localization and other techniques in the video below.