Is digital audio intrinsically "broken"?

The point with all these graphs are that jitter and other distortion products at the -100dB level or better.
Two graphs <> all graphs. If that were the case, we would measure two pieces of audio equipment and stop there. I post exact same measurements a few pages back where distortion products were in -60 to -70 db range.
 
Two graphs <> all graphs. If that were the case, we would measure two pieces of audio equipment and stop there. I post exact same measurements a few pages back where distortion products were in -60 to -70 db range.
If you were referring to the "faulty" TI BurrBrown USB DAC, that's correct. More extensive measurements demonstrated that the distortion levels were too high, so the chip had to redesigned to fix the problem ...

Frank
 
If you were referring to the "faulty" TI BurrBrown USB DAC, that's correct. More extensive measurements demonstrated that the distortion levels were too high, so the chip had to redesigned to fix the problem ...

Frank
No, I am talking about this one:

1210Halfig2.jpg


It is not less than 100 db is it?

And some other random ones:

Outlaw:
306FUpFig12.jpg


Arcam:
http://www.stereophile.com/images/archivesart/98Mcdfig8.jpg[/iimg]

Burmester DAC:
[img]http://www.stereophile.com/images/archivesart/1299burfig10.jpg

Chord:
Chord64fig11.jpg


(older) Oppo DVD player:
507Oppfig09.jpg


I could go on. But to be far, lots and lots of devices now achieve excellent low jitter performance when measured out of their internal DAC. And many external DACs do superbly. None of this invalidates the original point: that this architecture makes it non-trivial to get it to perform.
 
Regarding dynamic range, don't forget the peak-to-average range runs to ~30 dB for movies, and around 17 dB for music (higher for some styles, lower for others). Recordings must take this into account to avoid clipping the ADC (and thus DAC on the other end). So, if the mastering engineer reserves say 20 dB for transient headroom, then your 16-bit noise floor of maybe 100 dB drops to 80 dB. Now consider the average sound level is perhaps 20 - 40 dB below that, and a 100 dB spur floor does not seem quite so extreme.
 
No, I am talking about this one:

1210Halfig2.jpg


It is not less than 100 db is it?
Hmmm, that seems to relate to an RME soundcard, and I couldn't find that image on Stereophile's website, I note other jitter measurement images for RME cards there look very good.

And some other random ones:
The Outlaw one appears to be the only one that goes higher than about -90dB, the latter figure I would still be comfortable with from the point of audibility.

I could go on. But to be far, lots and lots of devices now achieve excellent low jitter performance when measured out of their internal DAC. And many external DACs do superbly. None of this invalidates the original point: that this architecture makes it non-trivial to get it to perform.
I agree it's non-trivial. But I have always said that getting good digital sound is hard: it's only that I wouldn't point my finger very specifically at jitter being the root cause of the problem! Of course, if you improve the components in non-trivial ways in an effort to improve jitter performance, it's extremely likely as a side benefit that other problem areas will have been addressed, like less than optimum power supplies, and the SQ overall will benefit ...

Frank
 
Regarding dynamic range, don't forget the peak-to-average range runs to ~30 dB for movies, and around 17 dB for music (higher for some styles, lower for others). Recordings must take this into account to avoid clipping the ADC (and thus DAC on the other end). So, if the mastering engineer reserves say 20 dB for transient headroom, then your 16-bit noise floor of maybe 100 dB drops to 80 dB. Now consider the average sound level is perhaps 20 - 40 dB below that, and a 100 dB spur floor does not seem quite so extreme.
I agree with the ADC side, but is there any reason why in mastering the CD track that the volumes wouldn't be normalised, or close to it? I can't imagine any digital album available as a commercial product would have a maximum peak 20dB down, these days a lot of albums spend most of their time at 0 to -3dB down!

Actually, that reminds of a question I was going to bring up at some time, regarding levels fed to the ADC while recording, perhaps you or Bruce or someone can answer this: is there any inherent reason, or perhaps it's already done in some equipment, why the analogue feed to the converter can't be split to go to, say, 2 converters, with a level difference of say, 20dB difference between the two? The reason for this, one converter gets a much higher signal level, it's working in the sweet spot of its range, but now and again it will clip, deliberately so, but the other converter has 20dB headroom and picks the signal up cleanly during this clipping period -- both converters record the session to 2 separate sets of tracks. Then, after the session, the clipped areas of the higher quality recording are repaired by the "safety" recording, trivially easy to do for software, and instantly you have 20dB better dynamic range for recording.

Thanks,
Frank
 
why the analogue feed to the converter can't be split to go to, say, 2 converters, with a level difference of say, 20dB difference between the two? The reason for this, one converter gets a much higher signal level, it's working in the sweet spot of its range, but now and again it will clip, deliberately so, but the other converter has 20dB headroom and picks the signal up cleanly during this clipping period -- both converters record the session to 2 separate sets of tracks. Then, after the session, the clipped areas of the higher quality recording are repaired by the "safety" recording, trivially easy to do for software, and instantly you have 20dB better dynamic range for recording.

Thanks,
Frank
A less crude version of this was implemented in HDCD to provide 20 bits of resolution in a backward compatible format to 16-bit CD. The format shifted the DAC up and down to adjust to the range that it needed to play.
 
I agree with the ADC side, but is there any reason why in mastering the CD track that the volumes wouldn't be normalised, or close to it? I can't imagine any digital album available as a commercial product would have a maximum peak 20dB down, these days a lot of albums spend most of their time at 0 to -3dB down!

Actually, that reminds of a question I was going to bring up at some time, regarding levels fed to the ADC while recording, perhaps you or Bruce or someone can answer this: is there any inherent reason, or perhaps it's already done in some equipment, why the analogue feed to the converter can't be split to go to, say, 2 converters, with a level difference of say, 20dB difference between the two? The reason for this, one converter gets a much higher signal level, it's working in the sweet spot of its range, but now and again it will clip, deliberately so, but the other converter has 20dB headroom and picks the signal up cleanly during this clipping period -- both converters record the session to 2 separate sets of tracks. Then, after the session, the clipped areas of the higher quality recording are repaired by the "safety" recording, trivially easy to do for software, and instantly you have 20dB better dynamic range for recording.

Thanks,
Frank

The peaks are near full-scale, but the average level of the source itself is much lower than the peaks. I do find it very hard to believe any music is compressed to the point that the average level is only 3 dB down.

Getting the two independent precision ADCs aligned and tracking (stably) on lsb boundaries is a nightmare. And, at any given technology node you can't do much about the noise floor. Still, it has been done, but high-performance ADCs offering 16 to 24 bits of resolution and noise floors well below -100 dBFS have obviated the need for "stacked" schemes.
 
The peaks are near full-scale, but the average level of the source itself is much lower than the peaks. I do find it very hard to believe any music is compressed to the point that the average level is only 3 dB down.

Getting the two independent precision ADCs aligned and tracking (stably) on lsb boundaries is a nightmare. And, at any given technology node you can't do much about the noise floor. Still, it has been done, but high-performance ADCs offering 16 to 24 bits of resolution and noise floors well below -100 dBFS have obviated the need for "stacked" schemes.
I misunderstood you about the average vs. peaks, but regarding compression there is some pretty crazy music out there: one chap tried the most extreme level of compression on a commercial album, and only was able to squeeze another 1dB out of it. There is not too much difference in average levels between these albums and full power square waves at certain frequencies, bizarre stuff !!

Regarding dual ADCs, I felt someone must have tried it at some stage, so it was good to get confirmation -- thanks for the info ...

Frank
 
Last edited:
Orb, the only thing that matters is the effective, absolute distortion level: if a -60dB signal has 1% distortion that sounds terrible to say, but 1% is -40db, add that to to -60dB signal, gives you -100dB actual sound level: inaudible.

Try testing a vinyl signal at -60dB: if you're lucky you'll get 100 - 1000% distortion and noise, good thing LP distortion is inaudible on a good system ... :)

Frank

So Frank,
tell me what happens with digital files with a mean -40dbfs, and music energy down to -110dbfs.
Such an album was reviewed in Hifinews this month, and I have given examples recently of other files with audible energy down to 100dbfs due to the mastering/recording, even many good recordings go down -80dbfs.
Furthermore, look at the Yamaha music server/player example.
The skirt stops existing at -100db, and yet several reviewers identified the same aspect that I quoted Paul Miller picking up in his measurements.

Going to stop there as I do not feel we will get any concensus on this topic :)
Edit:
I just read Don's comment and ties in with exactly what he says, so leaving it at this.... for now :)
And spot on Amir.
Cheers
Orb
 
So Frank,
tell me what happens with digital files with a mean -40dbfs, and music energy down to -110dbfs.
Such an album was reviewed in Hifinews this month, and I have given examples recently of other files with audible energy down to 100dbfs due to the mastering/recording, even many good recordings go down -80dbfs.
Because I say that the sound is inaudible doesn't mean it's not there, but you'll have to go to some effort to hear some musical sound or otherwise that exists at such a low level. You will need to turn up your system to maximum volume and stick your head right next to the midrange or driver, being extremely, extremely careful to be in an area of the track where there is absolutely no normal music or sound energy. If the digital track happens to be remastered from analogue tape you will be deafened by the tape hiss for a start, that will be at least 30 to 40dB louder than this very low level sound.

People need to do experiments with normal music tracks that have been attenuated, reduced in volume by 60 or 80dB: you'll get quite a shock at how extremely quiet they are. Especially the 80dB reduction, you should get an award if you can get even close to identifying that something is happening ...

Frank
 
People need to do experiments with normal music tracks that have been attenuated, reduced in volume by 60 or 80dB: you'll get quite a shock at how extremely quiet they are. Especially the 80dB reduction, you should get an award if you can get even close to identifying that something is happening ...

Frank
If I am playing an instrument and there are room reverbrations in the notes recorded that give us a sense of space, what do you think the levels of those reverbs are? Or when a note decays into silence. How low do you think that gets?
 
If I am playing an instrument and there are room reverbrations in the notes recorded that give us a sense of space, what do you think the levels of those reverbs are? Or when a note decays into silence. How low do you think that gets?
Not as low as -90dB, that's for certain. I know you'll hate me saying this, but go back to good ol' R2R. No-one complains about tape not being able to have a note decay to silence, but the signal to noise ratio on this medium is of the order of 70dB, at best. Which means that if you can hear anything softer than 70dB then you will also be listening to constant tape noise, hiss. And I don't hear too many people complaining about this ...

Frank
 
The modern dematerialized digital music collection is a marvel—weightless, information-rich, strongly interconnected. Mine has nearly 1,000 full-length titles, more than 10,000 individual songs/tracks, over 750 GB of uncompressed audio, all automatically indexed and available to me with constant-time access. It is both the present and the future of playing music.

Still, people like vinyl. (I count myself among their number. Roughly 30% of my library titles were originally acquired on vinyl.)

Analog and digital unite when you ephemeralize your vinyl LPs—give them one last victory spin to change them from physical objects to cloud-friendly digital bits. Why? You can stop the aging process (for your LPs, that is). In some cases, you can reverse old damage. And you bring all the advantages of dematerialized play to your LP-entrapped content while preserving the unique sonic quality of vinyl.

High quality copying is possible. Gigahertz computer clocks facilitate high sampling rates. Near-zero cost of storage makes compression unnecessary. And the software needed to accomplish this is (almost) all free.

Learn more ...
 
High quality copying is possible. Gigahertz computer clocks facilitate high sampling rates. Near-zero cost of storage makes compression unnecessary. And the software needed to accomplish this is (almost) all free.

Great idea --- But to digitize my approximate 1000 record albums is no small task since I must be in the room to flip records, clean records, change records, etc. Under normal circumstances and given the amount of free time I have available to just listen to music, that would take a VERY, VERY, VERY long time. And the time it takes me to do that, I would not be listening to the approximate 450 CD's I have yet to hear, scan and listen to some of the other 700 CD's I already have listened to (Or watch any of the 500 movies I own or ...... )

So much music ..... so little time!

Maybe there is money for someone to get into the high quality LP digitizing business !!

I am actively considering getting out of analog (LP's) for this very reason. I simply don't have the time to listen to and/or watch just the digital stuff I already own. But if I could get all of my LP's digitized first, that would be very very nice!
 
No doubt. Transferring all that analog content will consume a lot of time and TB of storage. The former is precious, the latter is essentially free.

Recording is real-time only. But you only have to go through it once for each title. I recommend archiving that recording as "original" and keeping a "working" copy to play with.

You can process (and re-process) the original any time you get new tools, learn new tricks, or need some new output created. Some of the tasks in the pipeline grow linearly with the audio running time; others grow linearly with the number of individual "tracks" being created.

I've done about 250 titles over the last 18 months or so, and find the process quite satisfying. I have particularly enjoyed re-connecting my father-in-law with his collection of jazz albums from the 1950s.


TGD
 
One can use the approach I did with my CDs: start with the LPs you like the most. Then as you find time, you can digitize the rest. I used to use my forum time to rip my CDs :).
 

About us

  • What’s Best Forum is THE forum for high end audio, product reviews, advice and sharing experiences on the best of everything else. This is THE place where audiophiles and audio companies discuss vintage, contemporary and new audio products, music servers, music streamers, computer audio, digital-to-analog converters, turntables, phono stages, cartridges, reel-to-reel tape machines, speakers, headphones and tube and solid-state amplification. Founded in 2010 What’s Best Forum invites intelligent and courteous people of all interests and backgrounds to describe and discuss the best of everything. From beginners to life-long hobbyists to industry professionals, we enjoy learning about new things and meeting new people, and participating in spirited debates.

Quick Navigation

User Menu