Are our CD transports really bit perfect? Can we trust blindly?

[Talisman]

Reading some threads about CD transports and the possible differences, I was absolutely certain that the digital output of any CD player was to be considered bit perfect, but I accidentally stumbled upon an article which questioned my beliefs.
The article has a scientific imprint, with measurements and a logical approach to the question.
Perhaps not all CD transports actually "sound" the same.

https://addictedtoaudio.co.nz/blogs...ll-a-cd-player-with-digital-output-do-the-job

 

[solderdude]

The signal going into a DAC may not adhere to certain standards which can potentially lead to audible issues in the DAC depending on the output of the transport.
This, however, is a DAC compatibility issue or a digital signal 'fidelity' issue not a 'sound' being added or bits not being the same as they should be (bit perfect) and also not 'interpolated' or processed.
Maybe there are some transports cheap DVD players) that add signal processing of sorts (upsample or whatnot) but those are not your standard transports in that case.

The digital out from a transport should just be the read sample values converted to a format the DAC can process.
There should be no alterations in sample rate nor sample values.
The 'bits' on the disc are not the actual sample values.

 

[Blumlein 88]

If want to test for bit perfect conditions do so digitally first off as doing analog measures of the output puts an extra point to obscure results in the middle. Clearly a device which does a 16 bit to 20 bit conversion is not bit perfect in any case.

I've tested several CD player digital outs, plus a couple DVD players and blu-ray players. All were bit perfect. Does not mean all in the world are, but pretty sure nearly all will be.

Beyond that when going over SPDIF/Toslink the DAC has to lock onto and derive the clock which can vary some between DAC/Transport combinations. That can cause very low level differences in the analog output though for the most part those differences are inaudibly different.

 

[Talisman]

The digital out from a transport should just be the read sample values converted to a format the DAC can process.
There should be no alterations in sample rate nor sample values..

I totally agree with you, and here's the point, I would have assumed that the digital signal coming from that cd player was bit perfect, because that's what we expect from a digital output, but hand measurements it seems that not yes and that the signal has undergone changes.

Maybe there are some transports cheap DVD players) that add signal processing of sorts (upsample or whatnot) but those are not your standard transports in that case.

The cd transport he mentions appears to be a mid-range dedicated cd player (not a cheap dvd player)

 

[Talisman]

I've tested several CD player digital outs, plus a couple DVD players and blu-ray players. All were bit perfect.

This is a very interesting data, the cd player indicated in the article could therefore be an exception (I hope it is)
However this will always make me doubt in the future, until now I had always considered absolutely sure that every digital output was bit perfect, evidently it is not.

 

[solderdude]

The cd transport he mentions appears to be a mid-range dedicated cd player (not a cheap dvd player)

The plot shown seems to indicate a DAC not properly reacting to its (SPDIF) input signal rather than data being changed.

 

[Talisman]

The plot shown seems to indicate a DAC not properly reacting to its (SPDIF) input signal rather than data being changed.

Do you assume that the reviewer misunderstood his measurement data? I don't have enough expertise to determine that.

@restorer-john @pma @amirm Could it be a reviewer's misjudgment?

 

[solderdude]

The lobing is typical of sync issues (jitter), most likely due to level or impedance matching issues.
That kind of signal is not caused by processing or signals not being bit perfect.

 

[Talisman]

The lobing is typical of sync issues (jitter), most likely due to level or impedance matching issues.
That kind of signal is not caused by processing.

What about the background noise floor instead? it clearly can't be lowered by jitter, and it doesn't make sense that it's at -115db on a 16-bit signal

 

[pkane]

The lobing is typical of sync issues (jitter), most likely due to level or impedance matching issues.
That kind of signal is not caused by processing or signals not being bit perfect.

It does look like either the measurement or the SPDIF input isn't synced properly. Measuring noise floor when the signal isn't properly received is useless. The device might just be cutting off output (a lot of DACs do that) when it can't sync to the input or can't sense a certain level of signal. That explains the lower noise floor below 16 bits, simply because no bits were actually passed through to the output.

 

[Cbdb2]

When any $30 CD rom can read your data bit perfect at 20 x speed I think any audio CD player can give you bit perfect at 1x speed. (unless the disc is damaged).

 

[DVDdoug]

You CAN get read errors and the drive makes a difference. If you rip CDs you can use AccurateRip to check for a bit-perfect rip and sometimes there are errors. It's almost always the CD but some drives are better than others at ripping marginal CDs. There is a statistical database for "computer drives" but I don't know if there is anything similar for audio CD players or "transports".

Most audio players are pretty good at error-correction/error-hiding. I'm sure you've noticed how rarely you hear a defect and I'm sure you've noticed that most CDs are fine and it's a problem with the particular bad/damaged CD.

In the end you're listening to analog out of the DAC so there are no "bits" and no "perfection".



... I just scanned-through the article, but

. If, for example, you have a 16-bit test signal which contains a section of “digital zero” for testing the noise floor, this may indeed to be resampled – from a 16-bit digital zero to a 20-bit digital zero.

As noted, the intrinsic noise floor for 16-bit audio is 96.3dB. For 20-bit audio it’s 120.4dB. That nicely encompasses the actual 115.4dB measured result.

In the digital domain, a zero is a zero, and a string of zeros is minus infinity dB. There is no noise except the analog noise that comes from the analog-side of the DAC (or whatever noise that exists in the recording). You don't get quantization noise until you have at-least one bit of data. If you've ever listened to an 8-bit file you can hear the quantization noise as a "fuzz" riding on top of the signal (at about -48dB). But when there is silence the quantization noise goes-away completely (minus infinity in the digital domain).

From what I've read jitter also sounds like noise but you can't hear it unless you somehow generate an unusually high amount of jitter.

Dither is also intentionally added noise and if the audio is dithered that usually exists even during silence (i.e. the zeros are not zeros) but a 16-bits or better you can't normally hear dither (or the lack of dither of the effects of dither.)