Airport Express/Streaming Audibility Tests

Airport Express - Does it or doesn't it corrupt the data stream? - the question seems to polarise opinion.

Seems to me we can do a simple test for this.
A couple of standard lossless tracks v the same tracks having passed through the AE.

All for download and blind comparison in the usual way.

Just sounding out the likely interest in this - if there is some I'll set things up.

 

How do you intend to capture it. There's little point recording the analogue out given it's low quality.

 

Apple lossless streamed from a Macbook Pro to the Express.
Optical output from the Express captured to hard disc via optical input on another Macbook Pro.

That v the straight lossless rip to HD.

 

That makes little sense IMO as you are only capturing 1s and 0s, and I'd be amazed if they were any different. The test would be more valid, especially given that you have a Cambridge 840 CD player, if you were to capture the analogue out whilst playing a CD, then capture the same analogue out whilst Wi-Fi streaming via the Airport Express to the optical input on the Cambridge, i.e. you'd be comparing transport mediums via the same DAC by assessing what actually comes out in the analogue domain. My guess, assuming the recording equipment is up to capturing it, is there will be a clear difference here. Certainly my own experience of comparing my MacBook connected optically to a dCS Elgar vs. the complex and expensive dCS CD transport showed the latter to be hugely superior. And no, I can't explain why...

Tony.

 

THe optical out from a Mac is poor, async firewire or async USB wil give you a far better result.
Keith.

 

But this is the confusing thing to me - surely that shouldn't matter when hooking it up to a upsampling DAC that I would assume would buffer and reclock the data. I'd be very surprised if the optical out compromised data at a bit / byte level.

Tony.

 

Rob that would be a good bit-perfect check, but because the macbook doesn't have to record the digital in realtime, there is no forceframe from a wordclock the process would effectively strip out any jitter in the signal anyway.

As to recording an airport express into a 840, good lock with even getting the 840 to lock onto the input, never mind hold onto the signal. Mine just chirps and burps whenever I tried an AE, and I tried 3 of them.

AE doesn't corrupt the data stream, but it does add jitter and the eye pattern is terrible, both of which affect the sound.

 

T Hi, well that is my experience, optical would be my last choice ,async USB EVO or async firewire Weiss INT would be my first.
Keith.

 

No longer have an 840 but it laways worked fine with AE.
I do have Cambridge Dacmagic and a Meridian Transport among others.

I don't see the issue with recording to the optical input on the mac. Surely that can be as real time as listening direct - I can record to fast SD card it it helps.
Whats the difference to using a dac with a buffer?

What I'm proposing we test here are the possible effects of jitter and the claimed lack of qulaity in the mac digital in/outputs. Putting the AE process and the mac interfaces inside the loop and comparing the result with a rip.

I can do analogue recording if people wish and think that of benefit.

That's what I meant by corruption - jitter primarily.

 

Funnily enough I already did some tests on this issue.

My focus was not on the absolute quality of the Airport Express, but rather the effect of driving it by iTunes or by AirFoil.

My DEQ acting as DAC will read out the exact sampling rate of the signal feeding it. It doesn't just show the nearest common sample rate as many do. As such I noticed that when driving the Airport by AirFoil, the sample rate actually drifted from 44.0KHz to 44.2KHz during playback. However, when driven directly by iTunes it was bang on 44.1KHz. I was curious to see if this was introducing massive jitter or other problems.

This was done using a Windows software, but the same is true for the Mac version of AirFoil.

So here we have the analog outputs of my Modified DEQ2496 being recorded and analysed when fed with different signals.


First up, the analog output of my modified DEQ2496 when fed by a direct optical output from a soundcard. I used a Behrigner UCA202, and before people claim it's not a good enough benchmark for the test, the exact same performance is seen from my more expensive M-Audio Fast Track Ultra digital output.

Jitter can be seen as the two side band spikes wither side of the 1KHz input tone.




Next up, this is the analog output of the modified DEQ2496 when driven by the Airport Express. The software is iTunes, where the sameple rate always reads rock steady.




Finally, this is the analog output of the modified DEQ2496 when driven by the Airport Express, with AirFoil. I used Spotify playing a local file to drive AirFoil.

*Please note here that the harmonic levels have changed slightly because for some reason Spotify outputs signals -1dB lower than other software. I should have adjusted for it by reducing the level of the other software but I didn't expect to be publishing this test data. The jitter should still be the same regardless of signal level.




My conclusion was that the performance of each method is equal, and it doesn't matter if I use iTunes or AirFoil, or indeed the Airport Express rather than a direct optical connection. the DEQ2496 doesn't re-sample by the way, it locks on to the input signal so it won't be hiding jitter problems.

 

Good stuff Simon.
I thought it might be good to attempt some audio comparisons to demonstrate any deterioration in performance. But I want a test that listeners have some faith in before we start.

 

Rob, the issue is that you have no idea what happens when you record via the optical in of the Mac. For all you know it could write all the data to ram and then produce the file. Or it could look at what data comes in per 1/44100th of a second and fill a data block whether anything comes in or not.

The issue is, you just don't know what it is doing, you have no idea what is really going on at a software level, and no control, so it's an utterly pointless exercise.


Simon, why do the sidebands either side of the 1khz test signal represent jitter? As opposed to some other distortion function?

 

That's never going to happen, is it? Some one will only have an opposing view (nothing wrong there) or have an agenda, etc. etc.

 

That's true, but then different dacs handle the incoming signal in different ways so there will never be a 'standard' in that respect. What concerns me is that the AE is very often described as 'crap' but I note that no such consideration for the interface is ever given when that harsh and negative comment is made.
There must surely be many instances were the AE output is buffered and/or reclocked yet I've never seen anyone report that the AE as an interface is transparent. It surely should be in many situations.

Perhaps recording the analogue output of a dac is the best way to proceed in this case. No probem if that is what people prefer.

 

Of course Dev, I fully expect that and some will always look for wriggle room in any test that might possibly challenge a particular view or interest, but I'm just looking to make the test as tight as possible.

I thought it might prove interesting but I'm not investing a few hours into doing this and uploading files if there isn't interest.

I've done it and have a firm view as a result. Others might not and are perhaps being guided by press and forum chatter on all matters digital. That's why I keep on throwing these trests out there, to challenge assumptions and force people into using (only) their ears. I have many others in the wings but I'd rather not be talking to myself

 

I'm not sure on the reasons jitter results in those side-bands, but other distortions are usually harmonic or intermodulation, both of which don't fit with what's happening.

Harmonics are multiples of the input signal and intermodulation occurs at either the sum of two frequencies or the difference between them.

 

Interesting that everything is way down below -108db.

Not exactly significant!

 

I'm not at all convinced that Simon's sidebands are jitter.

Robs test could use a (not that great?) DAC driven from a good CD transport and an AE, the output captured consistently.

Paul

 

I don't think they are jitter at all, if they are then there's something very wrong with his DAC for it to be so tightly correlated to the signal.

I've seen the Stereophile AE test, that tells me all I needed to know about the jitter from the device, and comes from someone suitably qualified in the scientific method and with the requisite electronics background and experience of carrying out the tests.

I have little to no faith in recording the analogue output of a recaptured digital signal to tell us anything about what went on earlier in the transmission path.

If I want to hear how shit the AE is I'll plug it into my network and hear it chirp with my DAC.

 

When you used the Airport Express and your CA 840 had trouble locking on, did you drive it from AirFoil?