Back in April we ran a poll here on The Citrix Blog to get your feedback on the quality of the audio in a sample recording. We didn’t provide much background at the time, since we didn’t want to bias the results. But now I’d like to tell you what this was all about.

What you were listening to was an audio sample being delivered at the Medium Quality setting in XenApp, representative of audio as it sounds without the benefit of the RAVE (Remote Audio and Video Extension) technology used in our SpeedScreen Multimedia Acceleration feature for DirectShow/DMO/MF streaming media playback (and currently being adapted for Flash). For most non-RAVE use cases, the Medium Quality setting provides the best balance between sound quality and bandwidth efficiency. We also offer a High Quality setting which leaves the audio uncompressed (bandwidth consumption is high) and a Low Quality setting -- which I really think we should have called “Low Bandwidth” since no one really asks for low quality -- so that anyone still on a dial-up connection can at least hear something even though the sound quality leaves a lot to be desired.

The results of the April survey are summarized in this chart:

What’s your impression of the audio quality in the sample audio clip? Results:

(9446 total votes)
Poor - If it’s my only option, then I cannot use it with my voice/audio application 2426 votes: 25%
Bearable - I can use it to communicate with others, but only if I have to 2185 votes: 23%
Acceptable - I can use this on a daily basis 2605 votes: 27%
Great - When can I have this? 2230 votes: 23%

I’m no statistician, but the distribution of ratings strikes me as unusual; not what I’d call a bell curve or “normal distribution”. There were roughly equal votes cast for each possible rating.

We asked the members of our Citrix Customer Advisory Community to evaluate the same audio sample. The distribution of results was a bit different (I don’t know why):

Not acceptable Marginally acceptable Acceptable Very Acceptable
How acceptable is the audio?

7 votes: 9% 21 votes: 27% 38 votes: 48% 13 votes: 16%

I think the good news from these two surveys is that at least 50% of listeners consider the audio quality at the Medium setting to be either acceptable or great. But of course the bad news is that for anywhere from 9% to 25% of us, Medium Quality is not usable.

So, what can you do with this information?

For one thing, now that you know that this is a sample of audio at Medium Quality, you can use this recording as a reference if you’re not quite sure whether you have Medium Quality audio properly configured on your system. Accessing the blog post from a locally installed browser, you can listen to the sample and use it as a reference to compare against the audio quality on your own XenApp system.

This information is also of interest if you’re thinking ahead to delivering VoIP-over-ICA. With the new audio mixing capabilities of Windows Server 2008, you may be considering publishing a softphone on XenApp and delivering VoIP-over-ICA using XenApp’s bidirectional audio capability. As the various softphone ISVs achieve Windows Server 2008 Terminal Services compatibility, it should be possible to deliver VoIP-over-ICA to users on the same LAN as the XenApp server (ICA does not yet support UDP so over a WAN or Internet connection packet loss could gum up real-time voice transmission). If you can’t spare the bandwidth for uncompressed High Quality bidirectional audio (roughly 1 Mbps), Medium Quality provides a good balance between sound quality and bandwidth efficiency (it uses about 40 kbps in each direction).

Remember, our audio codec doesn’t even come into play if you are watching a video or listening to streaming audio using the RAVE technology in SpeedScreen Multimedia Acceleration. With RAVE, no decoding is done on the server; the native compressed media stream is redirected to the client and decoded there. So with RAVE you can even deliver full HD audio, with much lower bandwidth consumption than uncompressed High Quality audio.

Do you have use cases where your users would really value higher quality audio but the bandwidth consumption of the current High Quality audio setting is too high? Then I’d like to hear your feedback on the research work we’ve done exploring a new audio codec technology -- see my blog on Project Ulysses. Your input will help us gauge the priority of bringing that to market.  

Derek Thorslund
Product Strategist, Multimedia Virtualization