Improved error robustness for predictive ultra low delay audio coding

This paper proposes a new method for improving the audio quality of predictive perceptual audio coding in the context of the Ultra Low Delay (ULD) coding scheme for real time applications. The commonly used auto-regressive (AR) signal model is leading to an IIR predictor in the decoder. For random access of the transmission as well as for transmission errors, a reset of the predictor states, in both encoder and decoder is used. The resets reduce the prediction performance and thus the SNR, especially during stationary signal parts, since the resulting noise peaks could become audible. This paper shows that using adaptive reset intervals, that are chosen according to psychoacoustic rules, improves the audio quality.