Image Processing Reference
The elementary stream just contains video. Somehow the audio must be added to it once
that has been encoded separately. The result of the audio-encoding process yields an ele-
mentary bit stream similar to that of a video-compression process. These must be com-
bined carefully so they remain synchronized.
The audio will very likely comprise several streams. This is because audio is played
back through several loudspeakers to give a stereo or surround-sound effect.
It is not very sensible to deliver the video and audio separately, one after the other,
because unless you intend to download the entire program, you will not have all the mate-
rial available to start playing until the last stream begins to arrive. It is very efficient in
terms of file-based storage but is no good for streaming.
The encoder postprocesses the compressed content by slicing it into smaller, unit sized
packets. These are then shuffled together like a pack of cards as shown in Figure 14-5, mak-
ing sure that packets belonging to the same real-world time frame are adjacent to each
Once these are interleaved, the stream is transmitted and at the receiving end, the
player separates the packets out again to reconstruct the elementary streams. The output
of this process is a collection of elementary streams that can be played together.
Once you have both the audio and the video encoded, some output targets will require
that the audio be embedded into the digital-video bit stream. This is commonplace in the
broadcast industry, as it is used to reduce the amount of additional cabling and routing
If some of your source material arrives this way, having flexible embedding and de-
embedding tools available will get you out of a hole. This falls under the general advice of
always being on the lookout for useful tools.
Flattening Out the Bit Rate (Traffic Shaping)
You may have controls in your coding process that allow you to alter the buffering and bit-
rate distribution during the encoding process. By carefully adjusting the buffering during
times of low bit rate, more frames are delivered to the buffer in the client receiver. When
some complex content is compressed, the encoder works within the available bit rate but
allows the client to run slightly ahead and catch up a little with the buffered stream. Then,
when the complex, high bit-rate content has been processed, the buffering begins to pick
Figure 14-5 Multiplexed packets.