One-Chip Voice Processing Platform Targets Hands-Free Comm Systems

The ZL38004 voice processing platform is a highly integrated, flexible, and programmable solution for hands-free communication systems. It includes dual-channel codecs, a Zarlink voice processor engine for acoustic echo cancellation and noise reduction, and complete firmware support for hands-free car kits (ZLS38500) and speakerphones (ZLS38501). The firmware helps designers deal with complex noise situations, such as echo, background noise, and double-talk (when two people speak at the same time) that have traditionally hindered hands-free systems.

To reduce noise in a car, the ZLS38500 firmware distinguishes between noise and voice signals, removing the noise and preserving the integrity of the original speech. Good double-talk performance relies on the interaction between the acoustic echo canceller (AEC) and the non-linear processor (NLP). The ZLS38500 firmware tracks changes in the echo path and continuously converges during double-talk and the NLP cancels the imperfections of the AEC.

The ZLS38501 speakerphone firmware is optimized for full-duplex operation and good double-talk performance. Most voice-processing solutions deliver only half-duplex operation, so that during double-talk only the signal with the highest level is transmitted. The ZLS38501 firmware continuously converges and tracks changes in the echo path to support full-duplex operation.