Bytes Per Packet Formula

Unanswered Question
Jul 29th, 2010
User Badges:

Dear Friends:

I would like any could help me whit this subject, because I am turning crazy!

In the bytes per packet formula, it says: bytesperpacket=(samplesize(s)*bandwith_codec(bps))/8.

For example, with codec G.729 (8000 bps), the result will be: (0.02s*8000)/8=20 bpp.

I don't understand this formula, and I believe that I have reasons for this:

If the sample has 20 ms (0.02 ms), and the bandwith of the codec is 8000 bps, I would take one sample every 20 ms, so, in one second, I would have 50 samples.

The codec has 8 bits for every sample, so for 50 samples, it would have 50*8=400 bits per packet, and if I divide it by 8 (absurd...) I have 50 bytes per packet, don't 20...

I believe I am having a fix between the cuantization and the compression, but I don't have clean the rol of the codec working in bps, I would understand better if it works saying how many bits compress the packet.

Thank you everybody.

Sorry my ignorance, but I am new in the voice world...



  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Chester Rieman Mon, 08/02/2010 - 09:49
User Badges:
  • Cisco Employee,

I think sample size may be throwing you off - it's actually 10ms per sample and we place 2 samples in one packet(that's where 20ms comes from.)

Hope this helps


This Discussion