Digital Audio and Amplifier Noise Floor Comparison - Is 16bit/44.1kHz All We Need ???

[sandyk]

1 minute ago, psjug said:

 

I'm not sure I am following you.  Are you saying it is nice to have extra amplifier dynamic range to account for level mismatches when paired with multiple sources?  I can buy that argument.

In part. Many sources such as Vinyl and Surround Sound, especially in live performances from DTV may  have channel levels well below that of the nominal 2V of RBCD

[Hugo9000]

Just now, psjug said:

But if you hear your own breathing, what does that mean?

  Well, I do think about my own death quite often, but I breathe as silently as I can, even when I'm alone haha!  I take special care with my breathing when I have a cold.  I really hate the sound of breathing!  hahaha!

[sandyk]

4 minutes ago, Hugo9000 said:

Isn't there a meme about hearing people breathe?  Like "if I can hear you breathing, I have fantasized about your death" or something?  Or is it chewing that bothers most people?  I personally don't want to hear chewing or breathing!

 

Sorry for the OT!

 You can often hear the breathing of many recording artists. An example of that is " Queen-Another One Bites the Dust" where with a good system you can even hear the spittle in Freddy Mercury's throat . (so I have been told )

[Hugo9000]

24 minutes ago, sandyk said:

 You can often hear the breathing of many recording artists. An example of that is " Queen-Another One Bites the Dust" where with a good system you can even hear the spittle in Freddy Mercury's throat . (so I have been told )

Yes, that's true.  I sometimes can hear breathing on close-mic'd violin recordings.  I don't mind when it's Gil Shaham, as he produces such a gorgeous tone from his violin.  I think he's the only one who doesn't bother me haha!

 

 

Edited to add:  It's just the occasional rapid intake of breath, not some awful heavy breathing!  I don't want to give the wrong impression of his recordings.

[fas42]

What is important is to have a electronics chain where the gain can be varied by a large amount with no deleterious effects, such as starting to hear hum or conventional circuit noise - albums have a huge range of average recording levels, and you don't want to be caught with not having enough gain to reproduce at a satisfying level, for the occasion.

 

Talking of breathing, there can be an issue with the how the miking of things like string quartets are done - I don't want to become more aware of the body movements of the player, than of the instrument ...

[sandyk]

27 minutes ago, psjug said:

But if you hear your own breathing, what does that mean?

 

 Perhaps due to Asthma/Bronchitis, Influenza, Sex ?

[sandyk]

12 minutes ago, fas42 said:

What is important is to have a electronics chain where the gain can be varied by a large amount with no deleterious effects, such as starting to hear hum or conventional circuit noise

 

 Assuming that your hearing isn't too aged, if you can hear hiss or hum from a Speaker with your ears close to it, then it WILL degrade low level material a little .

[esldude]

43 minutes ago, psjug said:

That's a really good point.  But do really think 0 dB is a realistic assumption.  What is the perceived level and FR of normal breathing, for example?

Don't know about breathing.  I suppose I could record some and see. 

 

Yes I do very much think 0 db SPL is a realistic assumption.  I don't have the files up now, but had a thread where I mixed in different levels of noise and let people listen.  The noise was bandlimited to 3-5 khz.  I asked people to listen to music and set a normal enjoyable volume.  Then listen to the tracks with the noise added.  Telling me where they no longer heard it. 

 

My assumption was most listening average levels are in the 75-80 db SPL range.  So reports generally were noise was no longer heard in the music when it was -70 or so db.  There was some variance, but not that much.  So noise in the 0 to 10 db SPL level seems to fit.  This also is congruent with work by Louis Fielder at Dolby Labs about levels of dynamic range needed for playback.  That is where I picked up on the idea.  If you think about it, other than hum, or running appliances like HVAC, you'll hear noise in that 3-5 khz range.  Tape hiss or surface noise on LP or similar.  

 

With dither CD can be effectively somewhere close to 120 db dynamic range.  So while not all CD is quiet enough, the medium itself can be.  

 

I could put up some files that are silence other than dither, and let people hear it normal and with various levels of gain to determine how much gain makes the basic noise floor of the medium audible in their system.  I could use TPDF and shaped also.  

[fas42]

23 minutes ago, sandyk said:

 

 Assuming that your hearing isn't too aged, if you can hear hiss or hum from a Speaker with your ears close to it, then it WILL degrade low level material a little .

 

Which ties in with what I'm looking for in system replay capability - with one's ears very close to a speaker driver, only inches away, do I detect the presence of a driver, or clues that it's a reproduction chain I'm listening to? The obvious giveaway is hum or hiss, but tweeter 'spit' is just as meaningful, in fact more so.

 

The presence of detectable levels of unwanted 'extra' does degrade low level material in the recording - and the close listening reveals it, very easily. It is possible to completely eliminate those artifacts, and then it's very clear that the 16/44.1 format is as good as necessary; play a full symphonic track that's been attenuated by 60dB, as the actual recording, at maximum gain of the rig. You will barely be able to hear it, with your ear hard up against the speaker surface ... yet, the music will still all be there, the full structure will be clear, with no obvious problems.

[psjug]

Keeping it simple - if playing a silent track from your highest level source (or maybe your preferred source) at full volume is not audible, doesn't that mean your amp has all the SNR that you need?

 

Doesn't address the 16 bit question, though.  Also, nothing wrong with wanting an amp that has much more than enough SNR.

[sandyk]

33 minutes ago, psjug said:

Keeping it simple - if playing a silent track from your highest level source (or maybe your preferred source) at full volume is not audible, doesn't that mean your amp has all the SNR that you need?

 

That will depend on  whether you are at the listening position or with your ears close to the speakers.

[esldude]

1 hour ago, sandyk said:

 

That will depend on  whether you are at the listening position or with your ears close to the speakers.

If your ears are close to the speaker, then it definitely is fine. 

If from the listening position, I'd say it is fine from the listening position.  

 

Ears only have about 60 db of instantaneous dynamic range anyway. 

[sandyk]

23 minutes ago, esldude said:

If from the listening position, I'd say it is fine from the listening position.  

I disagree.

 I have been able to hear a small but worthwhile improvement from the listening position after further reducing the already very low noise level contribution from a DC Offset Corrector, which wasn't audible from the listening position, and really only JUST visible at the Preamp's Output at maximum gain and a 1K resistor across the Input, with  a CRO at maximum gain PRECEDED by a Low Noise 10 x gain battery powered Preamplifier !

[fas42]

It's straightforward to assess whether the system can go "quiet enough" - usually most problems arise as one attempts to go closer to live listening levels - can one keep increasing the volume, and the SQ retain its integrity? This is where highly efficient speakers, horns, win bigtime - they have no trouble passing ambitious SPL tests, because no particular stress is being placed on the electronics. Most conventional rigs start to degrade, and lose the ability to create realistic sound levels as the gain is increased; they just end up sounding like another "hifi".

[esldude]

https://www.dropbox.com/s/yendxejquynko22/dither checks.zip?dl=0

 

If you want to hear dither download these two 40 second files. 

One is tpdf dither and the other is shaped dither.  16 bit. 

 

5 seconds of dither, a beep, then 5 seconds with 10 db of gain. etc etc. until you hear dither with 60 db of gain.   Should give you an idea if you have good margin of error in the noise floor of your system with 16 bit dithered files.  

 

Set volume to a comfortable or slightly loud listening level and see how much gain before you begin to hear it. 

[PeterSt]

No music playing (no attenuation anywhere either).

 

-3dBFS Into 3 Ohms (Single Ended).

 

The latter picture plays a 24 bit file. When 16 bits is played, 20dB of noise adds (I don't have a picture of that for this power amp measurement).

 

This won't be everybody's situation, but here it is so ...

 

 

 

 

[semente]

A thread on noise-shaping / dithering: https://pinkfishmedia.net/forum/threads/noise-shaping.198251/

[sandyk]

36 minutes ago, PeterSt said:

The latter picture plays a 24 bit file. When 16 bits is played, 20dB of noise adds (I don't have a picture of that for this power amp measurement).

 

 Probably a silly question, but what happens when you add Dither to a 24 bit file ?

[PeterSt]

17 minutes ago, sandyk said:

Probably a silly question, but what happens when you add Dither to a 24 bit file ?

 

It adds noise ?

But that won't be visible because that would be ~24dB more down for one bit. So with 4 bits of dither it would theoretically show.

 

In practice is maybe isn't so simple because more forces are at work. One of them is applying dither to a layer which is 24 dB down in the noise already, possibly isn't so useful (let alone the noise being at -80dB which would be a far more normal situation for most). What I "try" is applying the system noise itself as dither. This is nicely random noise (if all is right). Thus for example, when I'd foresee that more than 24 bits resolving is not really achievable, I'd still add 3 more bits to accomplish natural dither. This looks better to me than taking out 3 bits from the 24, to add dither.

The NOS2 indeed was designed with 27 bits for this reason. Not that we ever made it for real ... 😞

 

As you can see, my approach is a little different from what's common sense. I mean, while one may apply dither to hear deeper into the noise, I first avoid the noise so the remaining bits can really be perceived because they are genuinely there. Adding noise/dither again, is a bit (haha) against my religion now ...

 

 

 

And latter "below" was an other 6dB attenuated, not really showing a difference any more (only a little). So this resolves to 22 bits (a small tad more).

 

In case someone asks: this noise is not really smoothed out (an FFT analyser would be able to do that).

 

 

[sandyk]

Interesting.

Thanks Peter.

[esldude]

8 hours ago, Shadders said:

This is the discussion on whether Red Book CD (RBCD) is all we need for audio reproduction given that amplifiers may or may not have a lower noise floor.

I suppose another factor not yet discussed is what would you consider enough dynamic range?

 

A simplified idea is we don't want more than 120 db SPL because more is damaging.  If we accept my suggestion 0 db SPL is a good bottom number, then we need 120 db dynamic range.  A small number of amps, DACs and speakers could manage such a thing.   I believe most will be limited on the upper end by what their speakers can manage.  110 db SPL is going to be more common, and perhaps in smaller rooms even 105 db SPL will be fully satisfactory.  There are plenty of good speakers that can manage 105 db or a little more cleanly.    Any thoughts about this?

 

 

My next question would be are there recordings with this kind of dynamic range?  Theoretically it would be possible to get close.  I think in practice much of the processing and microphone limitations will mean you can't get to 120 db in most recordings.  I think you'll struggle to actually get more than 80 db in recordings honestly.  

[Solstice380]

9 hours ago, psjug said:

But if you hear your own breathing, what does that mean?

 

It means you are alive!  🤪

[PeterSt]

Maybe I don't know how to relate it to things said before, but I think this is of importance too:

 

Once you are of the stance that better no pre-amp be used and no other means of analogue attenuation as well, plus no in-DAC voltage control (for volume), then you better have some headroom for the digital attenuation.

Right ?

 

So when I started out with that (with a 24 bits DAC), the digital volume control explicitly shifted the range of usable bits of the 16 bits file from -0dBFS /-96dBFS to -48dBFS/-144dBFS, this giving a lossless attenuation range of 48dB.

(to you speaks the creator of the lossless volume control )

 

Later on this didn't really work for 100% (lossless) any more, because we started upsampling and each upsampling step requires an additional bit. Thus, with 2x upsampling 96 + 6 = -102 dB is required. This chops off a bit at the bottom end, when the 48dB of dynamic "volume" range would need to be maintained.

 

Let me add that DAC + poweramp + efficiency of speakers + room size and what you can bear for SPL, quite carefully need to be tuned to each other. So that too could be the task of the audio manufacturer (might he be able to put everything and all together).

There is no free lunch (or be sloppy) anywhere, once you really like to have things to the "ultimate".

 

[Paul R]

Perhaps all we need for playback, even ignoring headphones.

 

But a lot of folks, myself included, do not buy into “perfect sound forever” from 16/44.1. Certainly not for recording, and actually not for playback, since most 16/44.1 DACs are not playing that bit perfect, but are oversampling anyway.  

 

So, if the DAC is forced to oversample to sound good, using zero and hold or whatever scheme you choose, does it not make sense to just provide the DAC with a higher sample rate and avoid all the upsampling and SRC in the first place? Storage space and network transmission is a non-issue to most audiophiles in 2019, so that is far from a compelling reason to continue with 16/44.1k. 

 

And yes, current systems and technology may have limits, but that does not mean tomorrow’s tech will have the same limits. So it makes a lot of sense to save the music we have in the highest possible quality format. It makes even more sense too have multiple copies of that music spread around to help ensure its survival through the next fire. 

 

So no, 16/44.1 is not good enough. Unless of course, it is all that is available for some music I really love. Then yep. But of course, I will keep that music in its redbook form, not archive it as MP3 or some other 3 letter format.  

[psjug]

8 hours ago, sandyk said:

I disagree.

 I have been able to hear a small but worthwhile improvement from the listening position after further reducing the already very low noise level contribution from a DC Offset Corrector, which wasn't audible from the listening position, and really only JUST visible at the Preamp's Output at maximum gain and a 1K resistor across the Input, with  a CRO at maximum gain PRECEDED by a Low Noise 10 x gain battery powered Preamplifier !

Even though you were trying to reduce noise, and did so, how do you know that the improvement that you are hearing can be attributed to noise reduction, and not an improvement in something else like some form of distortion?