An audio DSP BelaSigna R261 introduced by On Semiconductor can improve the signal to noise ratio of voice on mobile phones by 30dB, claims the firm.
This product works in close-talk mode where 30dB of signal to noise improvement is entirely feasible, or conference mode where you can still see 15dB of improvement in signal to noise at 2m, although this really does depend on the environment.
Featuring a novel approach to removing mechanical, stationary and non-stationary noise, the chip preserves voice naturalness for greater speech intelligibility even when the talker is further away or not optimally aligned with microphones providing unmatched freedom of movement for end-users. Designed to be compatible with a wide range of codecs, baseband chips and microphones without the need for calibration, BelaSigna R261 is easy to integrate, improving manufacturers' speed to market.
Two microphones are required, and there are two modes: close-talk and conference (up to 6m range). The modes can simply be pin-selected from power-up to use the product immediately with generic settings, or if you want to get more sophisticated there are 400 parameters than can be set with a 16kbit EPROM over I2C.
This solution is an expert system. Inside, audio processing follows no simple algorithm and is a combination of techniques. It is aimed at anything with voice pick-up: handhelds, webcams, laptops, netbooks, intercoms and voice recorders, for example.
Power consumption is under 16mA at 1.8V, and supplies up to 3.3V can be used.
It connects directly into a digital microphone interface (DMIC) or into a baseband chip’s microphone inputs and uses simple omni-directional microphones.
"Production line tuning of the microphones is not required," said On.