By using redundant audio as defined by RFC 2198, it is possible to reduce the impact that lost packets have on VoIP audio quality. The idea is to send a primary encoding of normal quality along with a secondary encoding of lower quality within the same stream, but temporally delayed. An example might be to have G.711 as the primary encoding and G.723 as the secondary encoding.
If a packet is lost then the receiver waits until the audio for the missing time period arrives in the secondary encoding and then inserts that into the audio output. The resulting audio has a slightly higher end-to-end delay but far higher quality under poor network conditions than PLC (Packet Loss Replacement) can achieve. The implementation for end-points is a relatively simple layer between RTP and the audio codec, closely tied to the dejitter buffer. RTCP is not affected by this scheme.
As we move towards HD Voice, expect to see more interest in this little used mechanism; expectations of quality are rising while the underlying problem of packet loss remains unsolved for the majority of consumers.
Generic Forward Error Correction defined by RFC 2733 is a mechanism for correcting errors in the RTP header due to corruption or loss. This is a different approach that does not directly improve quality, however having the headers can be be helpful in handling loss more appropriately.
Valid8 Protocol Engine can help assess network, device and application performance by providing detailed metrics for audio, video and signaling. For more on performance, scale, and conformance testing for telecommunications networks, visit www.valid8.com