Abstract :
In wireless car handsfree communications, the far-end speech is played by the speaker, reflected inside the car cockpit and collected by the microphone. Therefore, the far-end voice is sent back to the far-end talker, who will hear his own voice, with a delay due to the wireless transmission delay, typically in the range of 200 ms. In the car environment, an other challenging part is the detection of double-talk periods. Because of the very low signal-to-noise ratios encountered in car environment, energy threshold based algorithms are not suitable for this kind of applications. In Texas Instruments, a program called WAVES (Wireless Audio and Voice Enhancement Solutions) has been launched in order to provide hardware and software solutions for advanced speech processing algorithms in a wireless environment. This program includes developments for acoustic echo cancellation, speech enhancement and speech recognition for wireless applications, in particular for handsfree systems. In the framework of this program, an efficient echo cancellation algorithm have been developed. This algorithm has been especially tuned for car environment. It is fulfilling ETSI GSM requirements on handsfree systems
