High Resolution Audio?

[RayDunzl]

I mean we could get 1.001khz and 1.002 khz more easily for exemple.

You could put 1.001 and 1.002 frequencies into separate channels.

Putting them into the same channel (or combining them in the air), gives a slow change in the amplitude of the resulting wave eventually cancelling completely, then growing again.

Using 1000, 1000.001, and the combination to illustrate:

It takes about 17 minutes to go through that cycle.

 

[edechamps]

Putting them into the same channel (or combining them in the air), gives a slow change in the amplitude of the resulting wave eventually cancelling completely, then growing again.

And for those wondering, that's called beating.

 

[RayDunzl]

For humans normally a 3% difference in two frequencies will be heard. So at 3000 hz that would be 9 hz like 3000 hz and 3009 hz.

3% of 3000Hz would be 90Hz

 

[RayDunzl]

And for those wondering, that's called beating.

I surprised myself one day, put 400Hz in the left channel, 405Hz in the right, and with headphones still heard beating.

No cancellation in the air, and none in the channels.

Only in my head.

JJ had an explanation for it,

 

[edechamps]

I surprised myself one day, put 400Hz in the left channel, 405Hz in the right, and with headphones still heard beating.

Same page: Binaural beats.

 

[Blumlein 88]

3% of 3000Hz would be 90Hz

Decimal mistake. .3% not 3%. Equivalent to about 5 cents on the musical scale. I fixed it in my previous post with a note it has been edited.

 

[Calexico]

Maybe we ear more precisely the differences the small variations when they are in the timbre and the resonance of instruments?
For exemple when tunning a guitar very small difference can make the whole résonnance sounding bad. When heared only on one string you won't notice the small difference in frequency compared to the good one.
When playing a chord you will notice it.
Then on high res clavecin could highlight better the better precision and timbre could sound more close to the reality.

 

[eliash]

Your original method was incorrect, but in fact I think it led to a more correct result than your current method, because you also overlooked the effect of reflections.

In a typical domestic room, the "critical distance" tends to be around 75-150cm from the speaker. This is the distance beyond which the SPL of reflected sounds is greater than the SPL of the direct sound.

In practical terms, this means that if a speaker produces 84dB at 1W/1m, in a real room it is likely to produce about the same SPL at any distance beyond 1m, too (of course this will be frequency-dependent*). So your original estimate of 108dB at the listening position for an 0dBfs input signal is likely to be closer to the reality than your revised figure of 101dB IMHO.

*For this reason, I suggest you do the same experiment with noise rather than a single 400Hz tone.

Thanks for the hint, fully agree and did so in the meantime.
In case you have read it above, there is a slight uncertainty about the level comparability of the sine wave and the CBNoise. I tried to adjust the CBN level to -20dB VU when creating it with Audacity, a sine wave at -20dB does not yield exactly that value when played, but probably around 3dB less. I checked that now with the external tc-electronic Clarity meter, it shows 3dB more (sliding 10s LUFS) for the CBNoise than for a -20dB sine wave. So I result in some 10dB SPL threshold sensivity for the CBNoise #5, compared to 5-9dB for the 400Hz sine wave (from yesterdays measurement), assuming a best guess 105dB max. SPL in both cases as a reference. The difference could be easily accountable for standing waves in the room...

 

[Calexico]

@RayDunzl
That means if a dac isn't precise enough and output 1.002khz and not 1.001khz that can lead to new modulations not present in the record.
Each frequency with a small imprecision will interact with each other in a different way than in the record.

 

[Krunok]

@RayDunzl
That means if a dac isn't precise enough and output 1.002 and not 1.001 that can lead to new modulations not present in the record.

I have never heard of digital volume control in ESS and AKM chips producing such artifacts.

 

[Calexico]

I have never heard of digital volume control in ESS and AKM chips producing such artifacts.

I talk about precision in frequency.
Ex: 1.345khz encoded and 1.398 decoded. I don't know the precision for that. Do we know it ?
There is nothing with volume in this.

 

[Krunok]

I talk about precision in frequency.
Ex: 1.345khz encoded and 1.398 decoded. I don't know the precision for that. Do we know it ?
There is nothing with volume in this.

So you're saying that truncation error in 32bit float calculation with volume control in DAC chips can cause a frequency shift?

 

[Calexico]

So you're saying that truncation error in 32bit float calculation with volume control in DAC chips can cause a frequency shift?

No i just wonder how precise are the dacs as no tests show it.
We sould for exemple make a multitone signal and see if every tone is at the good frequency or if there is a small imprecision in the frequency rendered.
Imprecision in the value of the frequency not the level.
Maybe higher samplerate give more precision for that.

 

[Krunok]

No i just wonder how precise are the dacs as no tests show it.
We sould for exemple make a multitone signal and see if every tone is at the good frequency or if there is a small imprecision in the frequency rendered.
Imprecision in the value of the frequency not the level.
Maybe higher samplerate give more precision for that.

Any frequency shift would be manifested as distortion and if DAC would be generating such frequency shifts you wouldn't possibly be able to correct them with higher sample rate.

 

[edechamps]

@RayDunzl
That means if a dac isn't precise enough and output 1.002khz and not 1.001khz that can lead to new modulations not present in the record.
Each frequency with a small imprecision will interact with each other in a different way than in the record.

It is true that a DAC clock will run slightly slower or slightly faster than, say, exactly 44.1 kHz in absolute terms. That has the effect of shifting the entire spectrum of the signal down or up in frequency. However - and this is why what you're describing is not a problem in practice - that affects the entire spectrum equally, not just a few frequencies, so the relative relationships between the various frequency components in the signal is still preserved and reproduced faithfully. As to the whole-spectrum shift itself, it's not audible in practice because the absolute pitch difference that we're talking about, for pretty much all DACs (even cheap ones), is well below human audibility thresholds.

 

[Calexico]

@edechamps @Blumlein 88
I think no dac can exploit all the data of high res files for the moment. So first we should find a true 24bit dac that has 24 bit precision until fs/2 that mean at least until 40khz.
Is there any dac that have 24 bit precision at 40khz??

 

[Calexico]

It is true that a DAC clock will run slightly slower or slightly faster than, say, exactly 44.1 kHz in absolute terms. That has the effect of shifting the entire spectrum of the signal down or up in frequency. However - and this is why what you're describing is not a problem in practice - that affects the entire spectrum as as a whole, not just a few frequencies, so the relative relationships between the various frequency components in the signal is preserved and reproduced faithfully. As to the whole-spectrum shift itself, it's not audible in practice because the absolute pitch difference that we're talking about, for pretty much all DACs (even cheap ones), is so small it is well below audibility thresholds.

Yes but is there a precision in the frequency ? Is this always the same for the whole frequency range ?

 

[eliash]

For humans normally a .3% difference in two frequencies will be heard. So at 3000 hz that would be 9 hz like 3000 hz and 3009 hz. Now beating between two tones at the same time can be heard closer together, but that is a different thing than you are asking. And CD can handle that with more precision than humans.

We know that .2 db level difference will corrupt blind testing. .1 db will not. So that is somewhere near the limit.

16 bit 44.1 khz can easily cover either one of these with a huge margin.

EDIT: I originally typed in 3%, but the proper number is .3% which is 9 hz for a 3000 hz tone.

I once build a passive headphone adapter to align various headphones´ levels to the speakers level. I asked myself how far the level steps could be separated and did some experiments. Whereas 1dB level difference were just distinguishable for me, 2dB were on the save side.
Maybe it depends on training, but I would not consider myself to be able to differ 0.2dB level difference...

 

[Calexico]

Any frequency shift would be manifested as distortion and if DAC would be generating such frequency shifts you wouldn't possibly be able to correct them with higher sample rate.

So does the actual test show this ? Is it proven that dacs do no frequency shifts?

 

[Blumlein 88]

@edechamps @Blumlein 88
I think no dac can exploit all the data of high res files for the moment. So first we should find a true 24bit dac that has 24 bit precision until fs/2 that mean at least until 40khz.
Is there any dac that have 24 bit precision at 40khz??

Actually there are DACs that are very precise even into the least significant bit. The problem is noise in the analog world. Around 20 bits worth of noise is what is possible. There are just a few that do a little better than that.