how tell if you're bit perfect?

[lewdog]

This is pretty similar to something I've asked in another forum, but more specific here...
How can I tell if a source I have is bit perfect (for redbook mostly, but others if I can tell)? Can I do some sort of test with some program or something? I don't think my sound card has an optical or coaxial in. I have a cheap dvd player that I'd like to know about because it's feeding a dac1. I know it's not a luxury player, but as long as all the bits are getting there I'd be sated.
Thanks.

 

[jefemeister]

It's pretty easy except you need a soundcard with a digital input, at least you do for my method.

The first step is to verify that the soundcard's recording capability is bit perfect. This can be done a few ways. The easiest is to take someone's word for it, but you can also design a statistical experiment.

To test you player you hook its digital output to the soundcard's digital input, make sure you have the input levels set flat. With cards like the 0404, you should set the clock to "external." You then make a recording using an audio editor. To get the reference file, use a program like EAC to rip the track directly off the CD.

Now you have to trim the tracks so they start and end in the same places. I look for the first non-zero sample on the left channel. I then trim the ending to an absolute time value, say 5 minutes. If you trimmed both tracks exactly the same you should now have two wav files that are identical (save some possible header info mismatch). The next step is to convert the wavs into a listing of sample values. I use GoldWave which has a utility to convert wav into a text file with ascii integer representation. You can then use a file comprison program to compare the two text files.

If there are any discrepencies found between the two files then you know the player is not bit perfect. If you don't find any, then you have only verified a higher probability that the player is bit perfect. You can increase the probability by repeating the test multiple times with different discs and using longer recording lengths.

Hope it helps.

 

[Glassman]

burn or get yourself DTS CD, hook your player to DTS supporting AV receiver, if it's sound what you hear, you're bit perfect, otherwise you'll experience just a noise.. that's a way when you don't have reliable digital input available..

 

[jefemeister]

Quote:

Originally Posted by Glassman burn or get yourself DTS CD, hook your player to DTS supporting AV receiver, if it's sound what you hear, you're bit perfect, otherwise you'll experience just a noise.. that's a way when you don't have reliable digital input available..

Yeah, I guess that would work too. Heck of a lot easier for sure

 

[lewdog]

Really? I don't understand the logic behind that (I don't disagree, I actually just don't understand). I actually am a bit intimidated by the first option, too...
Is it ridiculous of me to wonder if my source is bit-perfect? Do basically all sources do it right and am I just obsessing? I don't want to kill myself over this, but it does seem important (especially when spending large amounts of money) to do it right.
Thanks for your patience, guys. I really appreciate the help.
Lewis

 

[saint.panda]

Please excuse me if this has been asked before: Do bit-perfect transports sound the same if fed to the same DAC? Lewdog, I thought this questions might correlate to some if the issues you're interested in so please pardon the intrusion.

 

[sgrossklass]

Quote:

Originally Posted by saint.panda Please excuse me if this has been asked before: Do bit-perfect transports sound the same if fed to the same DAC?

Well, at least there's only the influence of jitter left then, which does reduce the variability a great deal.

 

[Don Quichotte]

You can use jefemeister's method until triming the two wav files, that is, until you get two theoretically identical wav files (in case your player outputs bit perfect data). Then you can use the wav editor included in EAC (tools -> process wav) to load one of the wav files, invert it (edit -> select all -> process selection -> invert), copy it (edit -> copy), then close this window and open another one loading the other wav (tools -> process wav); finally, mix paste (edit -> mix paste) the first wav over this second one and check the result. If the two wavs are identical, the result should be a straight horizontal line, regardless of how much you zoom in (some verification needs to be done). The process is simpler than it may appear, maybe except for the correct triming of the two wavs. There's also the "compare wavs" option in EAC (-> tools), but I'm not sure I understand how it works...

Quote:

Please excuse me if this has been asked before: Do bit-perfect transports sound the same if fed to the same DAC?

Saint Panda, by my knowledge the only other diference apart from the actual data (bits) should be the timing errors (jitter), so all the bit perfect transports should sound the same if fed into a 100% jitter immune Dac. Now, I don't know if such a thing exists (some say the Benchmark corresponds, but even more say it doesn't), but most Dacs are not jitter immune anyway, so for most of them the answer to your question is "No".

 

[saint.panda]

Thank you very much. I knew about jitter but thought bit perfect means jitter free, which admittingly was a bit hard to believe coming from a Chaintech AV 710 for instance. Again, thanks for the clarification!

 

[Cthulhu]

If you have jitter in your transport that audiably effects playback, wouldn't you get two different WAV files using the direct comparison method?

 

[Glassman]

getting exact data out through the S/PDIF is a piece of cake, the problem with not-bit-perfect sources is that someone screwed something, because nobody even cared what's going out when designing the device.. this can be a case with DVD players with their 'sound enhancing' DSP effects that are being applied to the output S/PDIF too, or with soundcards that can't work at 44.1kHz natively, or have bad drivers and such, but really, it's not a problem at all to get the data out correctly.. I'd say a majority of CD players have bit-perfect outputs, simply because they lack any digital volume controls or DSP effects so there is nothing to screw up.. jitter is what matters, jitter in audio area doesn't cause data integrity issues, but clock degradation, and that's audible in the end..

 

[jefemeister]

Quote:

Originally Posted by Cthulhu If you have jitter in your transport that audiably effects playback, wouldn't you get two different WAV files using the direct comparison method?

In order to get two different wavs, the jitter would have to be so great as to cause a bit error in the data. This would require an obscene amount of jitter to accomplish. Jitter is normally a much smaller timing problem. It simply causes the absolute edge of the sample clock to vary in location: ie, 44.1kHz +/- some very small percentage. This is not to imply that bit errors can not arise from other mechanisms.

 

[thomaspf]

The Benchmark DAC referred above is one of many models that use an asynchronous sample rate converter. In this type of design the DAC has a local clock and a variable set of incoming samples (dependinmg on jitter) is translated into a fixed set of samples in the local clock domain.

The net effect of this is that you will not find jitter induced side bands but you have the problem that every time you listen to a track your are listening to a different set of samples being converted. Different sources with differetn levels of jitter can result in very different samples being converted.

In the pure sense this is clearly not bit perfect. There is simply no way to play anything bit perfect on a Benchamrk Media DAC 1.

Now, can you actually hear the differences? .....

Cheers

Thomas

 

[clarke68]

I have two usb > s/pdif converters, one bit perfect (M-Audio Sonica w/ driver version 1.2.0.5) and the other not (Onkyo UD-5).

It's difficult to set them both up in a way that allows for accurate A/B comparison, but I seriously doubt I could tell the difference between them in such a test. I know for a fact that I can't tell the difference between them during normal listening. In fact, the only time it matters to me at all is when one of these 'bit-perfect' threads comes up.

Now, if you're hearing gaps, pops, or pauses in the music, then you have issues, but I've never heard of a properly functioning transport that did that except on discs that are badly scratched.

 

[Don Quichotte]

Thomaspf, I'm not sure I understand you correctly but I certainly wish I would. Do you mean that in a given time interval every transport outputs a different "quantity" of bits (or, to use a metaphor, reads the sentence above in, say, a second, from the very beginning to a varying point in it - one transport to the "y" in "certainly", another one to the "w" in "wish" etc.)? And this varying "quantity" of bits gets reclocked to a fixed "quantity" of bits, altering the content in this process? Is this possible, or I'm way off?