VoIP Voice and Fax Signal Processing

11.2 VOICE PAYLOAD AND HEADERS

A compressed voice frame is required to be packetized with Real-Time Transport Protocol (RTP), User Datagram Protocol (UDP), and IP headers and then encapsulated with network interface headers [Schulzrinne et al. (2003), Postel. (1980), URL (RFC791)].

The RTP header is 12 bytes. Voice is sensitive to delays. RTP helps proper end-to-end delivery of real-time voice traffic. RTP header compression reduces the number of bytes, but header compression is not considered in this book. Details on RTP are given in Chapter 10 and are available in RFC3550 [Schulzrinne et al. (2003)].

The UDP header is of 8 bytes as given in RFC0768 [Postel (1980)]. UDP offers a basic transport service of a 2-byte source port, 2-byte destination port, 2-byte length, and 2-byte checksum. The real-time aspects of transport are taken care of by RTP.

The IP header is 20 bytes in IP version 4 (IPv4) as given in RFC0791 [URL (RFC791)]. In IP version 6 (IPv6), the IP header is 40 bytes.

Compressed payload, RTP, UDP, and IP header combinations are called VoIP packets. A basic small block of samples used in voice compression is called a frame. In G.729A, 10 ms (80 samples) is the basic frame, and in G.723.1, 30 ms (240 samples) is the basic frame. Voice payload may use a group of compressed frames up to 80 ms. The group of frames is called VoIP voice raw payload or VoIP packet payload. An IP packet consisting of IP, UDP, RTP, and compressed voice payload is illustrated in Fig. 11.1. These ...