A voice activity detector for the ITU-t 8kbit/s speech coding standard g.729

Voice Activity Detectors (VAD's) are widely used in speech technology applications where available transmission or storage capacity is limited (e.g. mobile, DCME, etc.) and must be utilised with maximum economy. Modern day digital speech coding algorithms can provide toll quality speech at bit-rates as low as 8kbit/s (e.g. ITU-T G.729) and the use of a VAD can achieve further economy in average bit-rate. This paper presents a modified version of the GSM VAD, for use with the ITU-T 8kbit/s speech coding algorithm CS-ACELP, which makes an active/inactive decision for every 10 ms coding frame. The performance of the proposed voice activity detector is compared to that of the GSM coder in terms of VAD errors and subjective quality. Results indicate that the modified VAD has similar performance to the standardised GSM VAD while operating with G.729 parameters and coding frame size. 1. INTRODUCTION The use of a voice activity detector with the ITU-T 8kbit/s speech coding standard G.729 [1] converts the fixed bit-rate codec into a variable bit-rate version that is able to exploit the redundant silence exhibited by conversational speech. The ITU-T have recently defined a VAD algorithm [2] for G.729 that is optimised for simultaneous voice and data applications and is not specifically designed to deal with high levels of acoustical background noise. The aim of the work reported in this paper was to develop a VAD algorithm primarily for speech only applications where high acoustical background noise levels may frequently be encountered (e.g. mobile, speech over Asynchronous Transfer Mode (ATM) [3], etc). The VAD can be used in mobile radio applications to reduce the mean RF interference to other users or to reduce the power consumption of hand-held terminals. It can also be used in Digital Circuit Multiplication Equipment (DCME) [4] to detect the speech activity on the incoming trunks. Due to the need for robust operation in high levels of noise the VAD was based on the GSM voice activity detector [5], which works well, in conjunction with the GSM codec, under these conditions. The application of the GSM VAD directly to G.729 is not possible since it uses GSM coder parameters and threshold values based on the 20ms frame length of GSM. The modified VAD