Perceptual Audio Coding Using N-Channel Lattice Vector Quantization

Abstract

We consider the problem of reliable distribution of audio over packetswitched networks. We make use of multiple-description coding combined with transform coding in order to obtain robustness towards packet losses. Previous approaches to this problem were restricted to the case of only two descriptions. In this work we use nchannel multiple-description lattice vector quantizers (MD-LVQs), which allow for the possibility of using more than two descriptions. For a given packet-loss probability we find the number of descriptions and the bit allocation between transform coefficients which minimizes a perceptual distortion measure subject to an entropy constraint. The optimal quantizers are presented in closed form, thus avoiding any iterative quantizer design procedures. The theoretical results are verified with numerical computer simulations using audio signals and it is shown that in environments with excessive packet losses it is advantageous to use more than two descriptions. We verify in subjective listening tests that using more than two descriptions lead to signals of perceptually higher quality.

Low delay moving-horizon multiple-description audio coding forwireless hearing aids

“…An important difference to previous work see, e.g., [7,5] is that, in our case, we do not need the perceptual filter at the decoder. We do therefore not need to worry about whether smoothing the filter coefficients yields unstable inverse filters.…”

“…, n − 1} denotes the indices of the received descriptions. 4 In particular, the simple decoding rule where the reconstruction is given by the average of the received descriptions generally works well [10,5]. Thus, when 0 < m < n, we setŷ ℓ k = 1 m P j∈ℓ y j k , whereas when m = n we letŷ ℓ k = α −1 (y 0 k , .…”

“…These needed to be encoded and transmitted to the decoder as side information, in addition to the encoded audio data. It turns out that the bit rate required for encoding the perceptual model is up to 8 kbps [4,5]. Since this model is required in all the descriptions, the bit rate of the side information can be significant.…”

“…MD coding was recently used for robust perceptual audio coding [3,4,5]. In [3,4], the case of two descriptions was considered, whereas in [5] it was shown, that even with highly unreliable networks, it is possible to achieve audio streaming of acceptable quality by using more than two descriptions. In [4,5], perceptual models were derived at the encoder.…”

“…In [3,4], the case of two descriptions was considered, whereas in [5] it was shown, that even with highly unreliable networks, it is possible to achieve audio streaming of acceptable quality by using more than two descriptions. In [4,5], perceptual models were derived at the encoder. These needed to be encoded and transmitted to the decoder as side information, in addition to the encoded audio data.…”