A transform coding scheme for the transmission of audio data in ZigBee based wireless personal area networks (WPAN) is presented in this paper. Here, wavelet transform is exploited to encode the features of audio data included mainly in the low frequency region. As a result, it is confirmed that the presented scheme recovers the original audio signals much accurately while it transmits the binary data compressed as 37.5% of the entire data generated without coding scheme. Especially, the mean-squared error between the recovered and original audio data approaches $10^{-4}$ when the signal-to-noise power ratio is sufficiently high. Hence, the presented coding scheme which exploits the wavelet transform is possibly applied for high-quality audio data transmission services in a small-scale sensor network based on ZigBee. Such a result is considered to be applicable as a basic material to update the technical specifications and develop the applications of ZigBee in WPANs.