The XVF3000 voice processor provides far-field voice capture for smart speakers and conference calling applications.
The XVF3000 4-mic voice processor is designed for challenging acoustic environments. Acoustic echo cancellation (AEC) removes playback audio, enabling barge-in. Automated noise suppression and dereverberation remove unwanted noise and beamforming tracks and captures the voice of interest as it moves around a room.
You can parameterise and fine-tune the system to your product acoustics. With the choice of linear and circular mic-arrays, the XVF3000 is ideal for centre-of-room and edge-of-room conference-calling applications.
Acoustic Echo Cancellation (AEC)
Acoustic Echo Cancellation removes echo from the microphone audio input and enables the XVF3000 to detect voice signals even when high-volume audio is playing through the product, enabling barge-in across the audio stream.
The adaptive beamformer identifies the ‘Direction of Arrival’ and isolates the voice of interest. This can be read to the host processor or, in a conference calling application, converted into LEDs to indicate who is speaking.
Noise Suppression nulls stationary and non-stationary diffuse noise sources, for example air-conditioning and road noise where the frequency characteristics don’t change over time. This enables accurate, consistent voice detection.
Automatic Gain Control (AGC)
The Automatic Gain Control tunes the output channels for best results, whether that’s for an Automatic Speech Recognition Service (ASR) or communications.