Andy Abramson points out something I figured out a while ago: Skype audio quality is good, Gizmo Project's audio doesn't sound as good. I think that goes without saying. Many, many podcasters use Skype to record podcasts with far-away participants for one simple reason: it sounds better.
Why does Skype sound better? Quite simple: codecs. Codecs are algorithms that take voice or video data and convert it to ones and zeros in a compact way. There are many ways of converting voice (and voice) into ones and zeros. Some are well-known and free to use, others are proprietary.
Like many things Skype does, the codec they use is proprietary. The most recent versions use something called SVOPC. It is a wide-band codec that performs better than the G.722.2 wide-band standard. Both of these codecs sound better than your landline.
Gizmo Project does not appear to use any wide-band codecs at all. The end result? The call quality isn't anywhere near what it could be.
{ 3 } Comments
Skype’s audio has deteriorated quite a bit in the last years. Using it in 2004 it was a great experience, now I have like 1 out of 10 call that has acceptable call quality. Have they forgotten to upgrade their infrastructure?
Rich
Choice of codecs and how the media is handled between User Agents is an essential part of maintaining voice quality. We (at Truphone) have spent a very large amount of effort in getting this just right – consequently reviews of our audio quality out to PSTN, particularly to cellular phones is normally extremely favorable.
I used to illustrate this at speaking events by comparing Truphone audio quality with that of Skype – however, recently Skype upgraded their codec selection – they now use similar codecs to us – and the result is that the speech quality of calls (using the new client) to cellular phones is hugely improved! Well done Skype!
Codec selection makes a huge different, but a lot of things come into play. Do the endpoints support the codecs? Do the endpoints have enough horsepower to support those codecs? As packet loss increases, how well does the codec respond?
Post a Comment