10.8 ADAPTIVE JITTER BUFFER IMPLEMENTATION GUIDELINES

The following summary lists the most popular conditions and guidelines followed in designing AJB. This section is an expansion of Table 10.1. In a well-behaved end-to-end packet transmission, several design points will be relaxed in AJB and will behave similiarly to a fixed buffer.

  • Jitter buffer has to remove duplicate packets with minimal processing effort.
  • A dropped packet before reaching jitter buffer input cannot be recovered inside a jitter buffer. In case of redundancy, a forward error correction (FEC), or duplicate packets availability, algorithm has to maintain a best effort to substitute dropped packets. Catering to such substitutions may create extra end-to-end delay. Jitter buffer giving a dropped packet (meaning null packet to the decoder) can only be enhanced through PLC algorithm that is present as part of a voice chain decoder.
  • Jitter buffer on its own should minimize any additional packet drops.
  • Jitter buffer has to maintain a stationary pattern of speech and drop packets. If the drop duration is for 20 ms at the input of the jitter buffer, output should also maintain the same drop duration. It should not remove or insert extra silence/null packets to get the best out of PLC.
  • Most algorithms of AJB adjust packets on an as-needed basis. Any adjustment made in the middle of speech is harmful to the speech quality. As an improvement, jitter buffer should also look at the significance of speech conditions before ...

Get VoIP Voice and Fax Signal Processing now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.