Other people's experience with audio quality

Question

I was wondering what other people's experience has been with audio quality using P2P and/or Hub and spoke with LCCS.   We have a web based and an air based video chat client that we've been testing and so far our results are hit or miss (for both 2 way P2P and 3 way H&S).   We've been testing with hardware AEC, and without using headphones, with internal and external mics but I still get a lot of chopiness and it can be difficult to undertand each other even in a two way P2P call.
I know mileage will vary depending on how good the network connection is, but we're also using gmail's video chat right after as a control and the difference is night and day ( I know they have better audio codecs with variable bit rates and other fancy things etc etc.).  To be fair sometimes it sounds great even across country, but most times it's tough to understand one another.
Just wondering if everyone else was having similar problems or perhaps I'm missing something important here.  Any feedback is welcome.
Thanks,
-Eric

Answer 1

We are also running in a live site with mostly US to US participants although we have done early trials with clients in the UK as well. We also use the platform internally to host our team standup meetings and it works relatively well as long as everyone has reasonable bandwidth available to support all of the streams and everyone if wearing a headset to avoid echo...
We set up standard values for video and mic settings that we think work well for different connection speeds as well as expose both networking/bandwidth statistics as well as mic/video controls for making fine adjustments if an endpoint is experiencing quality issues. It all seems to work fairly well - if only we could get some sort of AEC in place then we could ditch the headphones and make it a whole lot better...
settings that we adjust on Camera include keyframeInterval and quality
settings that we adjust on Audio include gain, encodeQuality, framesPerPacket, silenceLevel, and echoSuppression
additionally we expose for monitoring network connection type, lccs connection type, available stream bandwidth, current video bandwidth, video frames dropped, audio stream bandwidth and audio frames dropped. These values are from the audio and video netStreamInfo objects as well as the connectSession streamManager and userManager (to get the connection type from a user's userDescriptor).
You should be able to adjust the various settings based on the user's connection type in order to optimize individual performance. I seem to remember that there was some sample that showed most of this, but I don't remember where that was...
-Chris