EMG signal compression with an ACELP coder using trellis-coded quantization of LSF parameters

Despite substantial interest in long term recordings of Electromyographic (EMG) signals, only a few studies deal with compression of these signals. In this paper, we develop a lossy coding technique for surface EMG signals based on the Adaptive Multi-Rate (AMR) coder and Block-Constrained Trellis-Coded Quantization (BC-TCQ) of the Line Spectrum Frequency (LSF) parameter. We show that the original waveforms of EMG signals can be preserved with a percentage error smaller than 5% using 1484.8 bits/s with a compression factor of 87.9%.