[shaocaholica]

Couldn't a lot of timing issues be solved with a packet based audio protocol instead of more expensive components with tighter tolerances?  Isn't that the engineering way?
 
By packet based I mean with send and receive buffers that can mitigate any error correction hiccups on poor quality data cables or EMI.  The receive buffer can also feed the DA converter on its own clock completely isolating the DAC clock from the USB source clock.
 
This is essentially how all streaming media works to account for data loss.
 
So you'd lose out on pure realtime audio but gain by being more more confident that there's no data loss all the way to the DA regardless of a lot of factors.

 

[xnor]

Audio streaming over USB is virtually error-free so I don't see a necessity for retransmissions and it already works by transferring packets containing several audio samples (1 ms worth of audio data for full-speed endpoints).
 
Streaming media usually does not retransmit either. It may however contain some redundancy for error correction if the communication channel is known to be unreliable.

 

[wakibaki]

As xnor says, nobody is really worried about data loss over USB.
 
What seems to cause the major concern is jitter.
 
The USB audio interface (the one that everybody uses, not the recent one) seems tailor-made to excite concern over jitter, or perhaps it's just that in the absence of any seeming concern over data loss, audiophile paranoia had to search out another cause for concern. Certainly the interface was designed without any awareness of the extreme stance likely to be taken by some audio enthusiasts and professionals (salesmen, not excluding sales engineers) and hence it is extremely awkward under conventional USB Audio Class 1 using Microsoft's driver to provide a stable, low-jitter clock at the player, which would have relieved any concerns about jitter. The clock must be recovered from the transmitted data.
 
The problem is that the player device has no control over the rate at which data is transmitted. This means that if a local clock is used, eventually a buffer over- or underrun will occur. This is addressed in the latest USB standard, but the latest standard is not supported by Microsoft, so that although USB Audio Class 2 has native drivers for Linux and MAC OS's, Class 2 Audio DACs remain (prohibitively IMO) expensive (especially since nobody has satisfactorily demonstrated that jitter does seriously affect devices using the previous standard).
 
It's a typical audio traffic accident, with all the usual suspects conspiring to ensure that the maximum fear, uncertainty and doubt are excited amongst customers. Good for business.
 
w

 

[shaocaholica]

I'm just reading up on the technicals of asynchronous usb audio and it seems to do everything I said in the op except for error correction and retry of frames.

I mean, if a frame is lost or damaged and no error correction or resends are taking place, shouldn't that produce a very audible artifact? Unless you hear those often, chances are there's nothing wrong with your cables/shielding/etc and you could get away with murder on cheap cables right? There's no nuances from one cable to another as the atomic effect of data compromise is a bad frame and that should be clearly and harshly detectable.

Someone correct me if I'm wrong here.

 

[xnor]

Well put wakibaki. FUD is indeed a powerful tool to control the masses.
 
 
shaocaholica, most of the time when there are drop-outs or audio glitches with a USB device it's neither the cable nor the device itself but the operating system of the host not (being able) providing the packets of audio data in time. After all a kernel routine needs to provide the next chunk every millisecond, or 1000 times a second.
In Windows the architecture is a bit flawed because other drivers (wifi, nic, graphics card ...) can block the USB streaming routine from being executed in time.
 
 
Back to the unlikely event of data corruption. Well yes and no. If 1 bit flips it could be the least significant which you probably won't notice, but if it's the most significant it will cause a huge change in the sample value.
1 => -32767
32767 => -1
16384 => -16384
 
This should cause a plainly audible *click*. If the data corruption is more severe you'll most likely hear very offensive bursts of noise or simply complete audio drop-out.
Btw, the error rate specified in the USB 2.0 spec is vanishingly low so unless you buy a crap cable (I'm not talking about price here but lack of certification) you should be fine.

 

[wakibaki]

If a frame were lost or damaged, then yes, there would be a noticeable effect, but the simple fact is that serial communications are very reliable and loss of data or error are very infrequent. Any cable rated for USB is rated for the standard quoted, USB 1, 2 or 3. Exotic cables are a waste of money.

w

 

[shaocaholica]

Ah good to know, thanks.  How can anyone say they can hear subtle differences in a usb cable with a straight face then?  Even from low and mid level vendors.  You'd think that kind of snake oil would be left for the top end.

 

[xnor]

  How can anyone say they can hear subtle differences in a usb cable with a straight face then?

 
Because they accept anecdotes, don't question claims or accept them in good faith, only "test" cables in a sighted fashion, ignore the power of bias and so on and on. When you confront believers with evidence it's not uncommon to see them sticking fingers in their ears going *lalalalalalaaaa* and making rationalizations why science can be dismissed in the case of audio components. I like this paragraph from Wikipedia:

Dissonance is aroused when people are confronted with information that is inconsistent with their beliefs. If the dissonance is not reduced by changing one's belief, the dissonance can result in restoring consonance through misperception, rejection or refutation of the information, seeking support from others who share the beliefs, and attempting to persuade others.

 

[wakibaki]

...it's not uncommon to see them sticking fingers in their ears going *lalalalalalaaaa*

lol. Well... silently to myself actually, but it conveys the sentiment.

w