Enhanced Packet Loss Concealment (EPLC)

Enhanced Packet Loss Concealment (EPLC)

? In the communication system, the bit error of the transmission link may be caused by the mobile terminal, interference between transmission links, or changes in the load of the TRX. When bit error occurs in voice transmission, at least one frame is lost before it arrives at the decoder, and thus the voice quality deteriorates

?To solve this problem, the Enhanced Packet Loss Concealment (EPLC) feature is introduced. EPLC predicts the information contained in the missing frame by using the information contained in the good frames and by considering the correlation between the previous frames and subsequent frames of the missing frame. In this way, the voice quality is improved.

? EPLC prevents the voice quality deterioration caused by frame loss and improves the voice quality. The effects of EPLC are better than the effects of the standard AMR algorithm stipulated in the 3GPP protocols. EPLC improves user experience, prolongs the average call duration, and helps increase the operators revenue.

?EPLC conceals packet loss by using Algebraic Code Excitation Linear Prediction (ACELP) and Huawei proprietary technique of waveform recovery. It aims to improve the AMR voice quality during the AMR codec process.

? EPLC extracts the voiceprint of the speaker from the information contained in the received good frames and stores the voiceprint. The stored voiceprint is then constantly fine-tuned based on the subsequent good frames. When frame loss occurs, EPLC predicts and recovers the missing frame using the stored voiceprint. EPLC is a complex packet loss concealment method for noticeable voice quality improvement.

* Voiceprint storage and fine-tuning

EPLC extracts the voiceprint from the good frames and store it. The extracted voiceprint comprises the LPC coefficient, pitch delay, pitch gain, and codebook gain. After subsequent good frames arrive, EPLC fine-tunes the stored voiceprint. This ensures that the stored voiceprint is faithful enough, as compared with the actual voiceprint of the speaker in terms of frequency, volume, and pitch

* Frame prediction and recovery

EPLC predicts and recovers the missing frame by using the stored voiceprint and considering the correlation between the previous frames and subsequent frames of the missing frame.

* Speech signal combination

The recovered frame is filtered and then combined with the good frames. In this way, the original PCM voice signals are obtained.