XMOS’ XVF3800 is a high-performance voice processor that incorporates state-of-the-art audio algorithms for the highest voice clarity in conferencing applications, which may face challenges of near- and far-field speaker intelligibility, echo feedback between input microphones and output monitoring systems, and room reverb, resonances, and background noise of the conference location.
XMOS’ reliable and cost-effective solution aims to deliver the auditory familiarity of face-to-face conversation to the remote working environment for Microsoft Teams and Zoom certification in conferencing devices, video bars, portable speakerphones, PCs, gaming monitors, and collaborative display solutions.
The XVF3800 processor incorporates four microphone inputs and allows engineers to fully optimise the acoustic performance of the end device, which will vary with enclosure and intended operating environment. This device comes in a 60-pin QFN package and supports USB 2.0 audio (and control), as well as Class 1.0 and I2S host audio interfaces at 16 kHz or 48 kHz sample rate, with PCM output via I2S to DAC at 16 kHz or 48 kHz.
Key features of the audio algorithms include:
- Acoustic echo cancellation (AEC) – removes far-end echoes picked up by microphones.
- Adaptive multiple beamformers – tracks and targets stationary and moving voice sources to enhance double talk and dereverberation performance.
- Noise suppression – detects and reduces diffuse ambient noise, e.g., air-conditioning or road noise.
- Automatic gain control (AGC) – provides consistent volume between voices at different distances from the device.
The XVF3800 and associated XK-VOICE-SQ66 development kit are available to early access customers through special enquiry; general availability is expected from Q2 2023. If you are interested in incorporating this technology into a commercial design, fill out the form below with details of your project, and ipXchange will put you in touch with XMOS for further information and collaboration.