There exist a variety of MDC schemes of different approaches for audio coding in past literature [8][14][7], with different complexity. For low complexity, we use the scheme described in [7] to generate the two multiple described streams at the sender. The basic idea is to quantize the even samples in finer resolution (e.g. PCM, 8 bits/sample) and the difference between adjacent even and odd samples in coarser resolution (e.g. ADPCM, 4 bits/sample), and then pack them in the packet of stream 1. For stream 2, we quantize even and odd samples in the other way (Fig. 1(a)).
Figure 1: Source encoding: a) MDC; b) single-stream with FEC
During transmission, if the packet from one stream (e.g. stream 1) is dropped over by link, or discarded at the receiver due to its late arrival, the chance is still good that the packet from stream 2 can be successfully received and played out, due to the independent characteristics of the two channels. As a result the odd samples of the source signal can be reconstructed in full resolution, while the even samples in coarser resolution. The overall SNR is degraded, but the quality is till better than that of receiving no packet.
In order to compare the performance of multi-stream with that of the traditional single-stream, we use FEC as described in [4] for single stream. To make fair comparison, in the FEC scheme, the source packet is coded with the same finer codec, and a secondary copy is coded with the same coarser codec, and carried by the subsequent packet (Fig. 1(b)). In this way, the payload data rate of two-stream and single-stream transmission schemes is identical, and MDC does not increase the data rate of the payload compared to single-stream FEC. However one should note the added overhead of packet headers as a result of transmitting multiple streams. With 30 ms packetization time, 8-bit PCM/ 4-bit ADPCM coded payload, and 40 bytes of RTP, UDP and IPv4 headers for each packet, the overall data rate of multi-streaming is increased by 11%, compared to single-stream FEC.