This document was generated from CDN thread
Created by: Donald Newman on 03-11-2010 04:42:00 PM
We've been asked to design a VC system capable of transmitting all the nuances of a musical performance. The client is the Music School at UGA who want to have a "master class" type format where the performer would be local and a remote listener would view, listen and critique the performer. Obviously frequency response, dynamic range, and minimal processing of the audio will be critical.
IP network bandwidth can be leased to ensure that doesn't become an issue. We can design the front-end (mics, space, DSP, etc.) to meet the needs but...
What about the basic audio channel provided by the CODECs? Other than turning off the CODEC's echo cancelling what other flexibility is available to minimize the artifacts created by CODECs themselves?
We have designed a number of automated VC classrooms but this is obviously a little different.
Thank-you for your input,
Donald Newman
CTL - Classroom Support
University of Georgia
706-542-3456
Subject: RE: minimizing audio processing performed by CODEC
Replied by: Bjorn Winsvold on 10-11-2010 11:18:45 AM
You should switch off the AGC in addition to the echo canceller. Are you going to transmit a stereo signal or mono? If mono is sufficient, you should use the AAC-LD 128kbps codec to get best audio quality. The AAC-LD codec will transmit the whole audible frequency range up to 20 kHz.
If you configure the endpoint as described above, the endpoint will do as little processing as possible - the only audio quality degradation will be introduced by the AAC-LD codec, which in fact is quite a good audio codec.
I will assume that the main degradation of audio quality you may experience will not be introduced by the codec, but rather by the external echo canceller or other DSP in the audio chain.
I hope this will help you.
Regards,
Bjørn Winsvold
Manager Audio - Personal Systems
Cisco Systems
