Cadence Tensilica HiFi DSPs get optimised Conexant voice/speech processing

May 07, 2015 // By Graham Prophet
Claiming that it can “dramatically” improve VoIP communication quality and automatic speech recognition system performance, Cadence Design Systems and Conexant Systems have announced that Conexant’s AudioSmart software has been optimised for the Cadence Tensilica HiFi digital signal processor (DSP) family.

Conexant’s AudioSmart software provides clear voice quality, eliminates noise from the microphone input signal during VoIP calls, and dramatically improves automatic speech recognition (ASR) system performance. The result is a superior audio experience for the end user, with excellent listening performance and voice quality in noisy, real-world conditions.

Conexant’s AudioSmart software is based on its proprietary Smart Source Pickup (SSP) noise suppression algorithm. Unlike traditional technologies, such as beamforming, SSP is a blind source separation technology that uses spatial representation of target speech and noise sources to reduce interference. For example, SSP is extremely effective at reducing keystroke noise distractions during conference calls and other computer-based voice communications. With SSP, speech and noise can arrive at any angle relative to the microphones; SSP even reduces noise when it comes from the same direction as desired speech, and when the speaker is up to five metres away from the target device. AudioSmart allows users to experience excellent voice quality independent of device orientation or the user's position (angle) relative to the device.

Conexant adds, “Our software greatly improves the performance of other technologies also running on the Tensilica core. Technologies that makes use of speech recognition, such as an always-on voice engine, can benefit from crystal clear audio facilitated by AudioSmart.”