A packet loss recovery technique with line spectral frequency modification in 3GPP EVS codec

This paper examines a method for controlling the energy of decoded signal at the recovery frame from a packet loss. Our observation unveiled that a packet loss before speech onset causes sudden increase in the amplitude of the decoded signal at the recovery frame when predictive quantization of line spectral frequency is used. To mitigate the artifact caused by the overshoot, a detector of the overshoot is proposed as well as a method that controls the amplitude of the decoded signal by adjusting distances of adjacent line spectral frequencies. This technology is implemented in the enhanced voice services (EVS) codec which is the latest 3GPP speech and audio codec standard.

[1]  F. Itakura,et al.  Spectral smoothing technique in PARCOR speech analysis-synthesis , 1978 .

[2]  Redwan Salami,et al.  Wideband Speech Coding Advances in VMR-WB Standard , 2007, IEEE Transactions on Audio, Speech, and Language Processing.

[3]  Methods for objective and subjective assessment of quality Perceptual evaluation of speech quality ( PESQ ) : An objective method for end-to-end speech quality assessment of narrow-band telephone networks and speech codecs , 2002 .

[4]  H. Tasaki,et al.  Spectral postfilter design based on LSP transformation , 1997, 1997 IEEE Workshop on Speech Coding for Telecommunications Proceedings. Back to Basics: Attacking Fundamental Problems in Speech Coding.

[5]  Michael Schnabel,et al.  Enhanced time domain packet loss concealment in switched speech/audio codec , 2015, 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[6]  F. Itakura.  Line spectrum representation of linear predictor coefficients of speech signals , 1975 .