Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[AndyU]

You'll probably also want to disable dithering if you're trying to go for that, otherwise there are no guarantees.
But who plays without any kind of DSP anyway, and if its only loudness and volume leveling.

I play without any kind of DSP! I predominantly listen to whole works of classical music for which volume levelling makes little sense - indeed, it's not applied to albums in any case. I have occasionally thought about using loudness, but I might be more inclined to do so if I had control over the level of loudness correction applied - I would want to tell MC to apply the (say) 80db loudness contour.  As it stands, MC doesn't know the volume level of my system, so I can't see how MC can know what level of loudness correction to apply. If it's not too much trouble, perhaps you could at least show the level being applied in the Audio Path window.

[Hendrik]

if I had control over the level of loudness correction applied - I would want to tell MC to apply the (say) 80db loudness contour.

Loudness works properly if you use internal volume as well as configure a volume reference level at which the zero point for loudness sits (ie. -20dB pink noise measured at 83dB at listening position)
So all you need for it to work is a volume meter to set the reference level - and of course use internal volume.

If you use an external volume control, which MC knows nothing about, then Loudness can't work properly.

But lets try to stick to the topic at hand, the new dithering algorithm!

[AndyU]

Loudness works properly if you use internal volume as well as configure a volume reference level at which the zero point for loudness sits (ie. -20dB pink noise measured at 83dB at listening position)
So all you need for it to work is a volume meter to set the reference level - and of course use internal volume.

If you use an external volume control, which MC knows nothing about, then Loudness can't work properly.

Thanks, I didn't know that. Just found details of how to do it too in the wiki here in case anyone else is interested. Still, tbh, not much interest to me, but I might give it a play - perhaps could be useful for the odd late night quiet listen. I wonder does BobKatz expect folk to use loudness controls for stuff he has mastered?

[mojave]

If I use ASIO then Audio Path reports 32 bit output. WASAPI on the other hand with Automatic Bit Depth set it in the WASAPI options then Audio Path reports 24 bit (padded) output. I appreciate there may be no audible difference especially in real life use but which one is technically superior?

If Audio Path ever reports 32-bit output, but you have a 24-bit DAC, then you need to go to Tools > Options > Audio > Audio Device > Device Settings > Output and check "Device uses only most significant 24-bits (Lynx, etc.)." This makes sure that JRiver dithers to 24-bits instead of 32-bits since your DAC will discard 8 bits after receiving the data.

Unless this is checked for ASIO in your case, then technically WASAPI is superior. By checking it, you make them both the same.

[kstuart]

 Supposing you start with a dithered 24 bit signal. You change the gain or equalize or whatever. This produces a 64-bit float output so you must dither to 24 bits again on the way to a 24-bit DAC.

So what does MC21 do with the 64-bit float output if one has a 32-bit DAC ?

[Hendrik]

So what does MC21 do with the 64-bit float output if one has a 32-bit DAC ?

It dithers it to 32-bit of course.
Assuming the original music had built-in dithering (which it really should), and you don't apply DSP, then the added dithering from MC will be far below the volume level of the built-in dithering, and completely irrelevant.

[glynor]

But of course, when I am listening to that album, I manually adjust the volume to be .... drum roll.... the volume at which I want to listen.

To each their own, but that's weird. Why would you do manually (and sloppily) what the computer can do more precisely and automatically for you? I suppose it is good for health reasons to stand up and turn the volume knob.  But I'd rather be able to switch to a different album if I change my mind and not have to adjust the volume (or risk getting blown out of my chair).

I think it is pretty darn weird to want to listen at "high-quality" and to you (and some others) this means with all of the cool features of MC that improve sound quality turned off. I'm with Hendrik. I can't imagine wanting to use MC without at least Volume Leveling turned on.

But, like I said, to each their own.

[mwillems]

It dithers it to 32-bit of course.
Assuming the original music had built-in dithering (which it really should), and you don't apply DSP, then the added dithering from MC will be far below the volume level of the built-in dithering, and completely irrelevant.

And with a 32-bit output the noise will be near or below -192dBFS which is several orders of magnitude below the effective resolution of any known DAC (to my knowledge, the DAC holding the record for signal to noise ratio has a resolution of about 140dB).  

[Arindelle]

Why would you do manually (and sloppily) what the computer can do more precisely and automatically for you?Why would you do manually (and sloppily) what the computer can do more precisely and automatically for you?

Well if I'm listening to an album I control my volume via JRiver without leveling.  I basically don't touch my-preamps volume control any more listening this way. Doesn't bother me as I'm manually choosing another album anyway if I have to adjust the volume.

Now I may be crazy, but I find my system sounds best when my analogue pre-amp is set to around 60 (on a scale of 100 -- I find the sweet spot for my amps is between 40 and 70.  If they are "underdriven" or if the volume is pushed up to compensate for lowered volume using Leveling, I just don't think it sounds as good (example putting the amps to 80 for VL,  sound level remaining the same)... am I totally nuts? Or is there some explanation why I would find this to sound better. At 58 I'm not hearing -120db or below noise floors thats for sure!

(for info I have two active crossovers and 2 amps - cross overs are between the pre-amp stage and the amps themselves - so an active bi-amped system I guess its called - so no digital x-overs. I also have a  crappy sub-woofer for high volume use, but I keep this off most of the time so that's not in the equation; the others are quality amps, but not too powerful, )

However, if I cue up a number of albums or run a playlist I do use volume leveling and am very happy with the results, just not quite as much. I'm not paying as close attention to the music so its doesn't matter though. And since the R128 changeover, I think this has really gotten so much better.

 sorry a bit off-topic, Bob

[glynor]

Well if I'm listening to an album I control my volume via JRiver without leveling.

I'm not sure how it is different if you control the volume yourself inside MC, versus using Volume leveling. Both are doing the same exact thing.  Now, you might think that, occasionally, you do a better job. But then, if it isn't quite right from the auto-adjustment, you can just tweak it by changing the internal volume, right?

That whole comment has the smell of placebo effect. That's fine, if it makes you happy.

To be clear, because it might not have come across.  My point wasn't that it was wrong to have it turned off.  At all.

But the tone of kstuart's comments throughout this thread, responding to questions from Hendrik, read to me like something approaching incredulity that everyone wouldn't obviously see things his way and do it the same way in his situation. There were a bunch of assumptions, and then a bunch of "Are you daft? Obviously I do it this way because of X, Y, and Z. " types of responses.

I was trying, and perhaps failing, to illustrate through allegory that it is not so simple. I, for one, fall in with Hendrik and I find it hard to believe that people would ever want to do it manually.  Sure, it might not always be perfect, but then just tweak it a bit.  You have to tweak it less often.  But, kstuart feels, apparently, quite the opposite.

That's fine. No one is wrong. I was trying to point out that it is reasonable to expect the opposite, despite the conditions kstuart presented.

[DarkSpace]

Just a small question that's been confusing me for a while now:

Dithering is never not performed as long as its enabled. But this is not bad. Even if you're playing 100% bitexact, then the signal ought to have baked-in dithering, or you run into the same old problems. If it does have baked in dithering, then re-dithering will just "overwrite" the original dithering. Dithering twice does not increase the amount of noise, it just overwrites the original noise.

This sounds like the following scenario:
Input is 16-bit 48 kHz, Output is 16-bit 48 kHz, Internal Volume is at 100% and all DSP effects (even Volume Leveling) are turned off.
As I understand it, this means that the 16-bit samples are converted to 64-bit float samples. Those 64-bit float samples contain only the original 16-bit data. Next, because all DSP effects are disabled and no Volume change occurs, the 64-bit float samples are converted to 16-bit.

In this scenario, there is no rounding error from the conversion of 64-bit float to 16-bit, because the 64-bit float samples are only the untouched 16-bit samples of the input. Does the dither that's applied in this scenario change the output samples or are they identical to the input samples?
I think that there should be no change to the samples, because as I understand the concept of dither, it is meant merely to compensate for rounding errors and this scenario has no rounding errors. However, I learned that about video processing, so it may well be different for audio processing. Also, the above-quoted post seems to imply that the output samples will be different from the input samples.

[Hendrik]

TPDF dithering is stronger than your typical video dithering, and as such it will also modify a signal if it has no rounding errors.
You can use JRiver Bit-exact dithering (ie. what we used up to now, and still as the default going forward) if you want to ensure such a chain remains bit-exact. It works more like the video dithering you had in mind.

Like I explained in an earlier post, trying to detect such a bit-exact condition and dynamically turning dithering off is rather complex and risky at best. I rather dither when not needed than accidentally disable dither when it would be needed.

Bit exactness aside, its important to remember that unless you severely reduce the volume in MC using Internal Volume (maybe to compensate for very strong amplifiers), even at 16-bit you will probably never hear the dithering noise, its still at -120dB or thereabouts. And once you do reduce the volume, you definitely want to dither anyway!

[kstuart]

But the tone of kstuart's comments throughout this thread, responding to questions from Hendrik, read to me like something approaching incredulity that everyone wouldn't obviously see things his way and do it the same way in his situation. There were a bunch of assumptions, and then a bunch of "Are you daft? Obviously I do it this way because of X, Y, and Z. " types of responses.

https://en.wikipedia.org/wiki/Attribution_bias#Hostile_attribution_bias

[AndyU]

..

I think it is pretty darn weird to want to listen at "high-quality" and to you (and some others) this means with all of the cool features of MC that improve sound quality turned off. I'm with Hendrik. I can't imagine wanting to use MC without at least Volume Leveling turned on.

..

Classical music has such a range, and dynamic range, that volume levelling between works would rarely give a useful result. Suppose you listen to a string quartet, or solo violin, after an orchestra. At what level should the second piece be played? As though the solo violinist were in the same concert hall as the orchestra - in which case very quietly, or as though the soloist were in the room with you, in which case much louder? It's an aesthetic decision that I prefer to make as a listener, and dependant on many things, including the style of the recording and it's content. It is most certainly not at all weird to prefer to make the choice myself.

[glynor]

https://en.wikipedia.org/wiki/Attribution_bias#Hostile_attribution_bias

Perhaps. It did seem like you were frustrated to have to explain yourself.  Explanation was obviously required, since there are plenty of legitimate "other ways to do it" was what I meant.

I certainly accept that it is difficult to read tone on the forum. I meant no offense and I assume you didn't either.

[glynor]

It is most certainly not at all weird to prefer to make the choice myself.

Yeah, but using volume leveling doesn't stop you from doing that. That's what's weird about it.  It gets you closer, is all.  You still have, you know, a volume control.

Like I said, it is fine, and you can turn it off. I don't think it is even on by default.

[JimH]

I'm going to let this go a little longer but close it soon. 

JRiver staff doesn't have time to be discussing all the details of dithering, and I personally think we've reached the very edge of what might possibly affect audio quality.  I don't want our development effort focused on all the things that "might" affect quality.  We need to draw a line and I think we're already a little bit past that line.

Or as we say in Minnesota, we have bigger fish to catch.

[DarkSpace]

TPDF dithering is stronger than your typical video dithering, and as such it will also modify a signal if it has no rounding errors.

Thanks, so that's where my misconception was.

Like I explained in an earlier post, trying to detect such a bit-exact condition and dynamically turning dithering off is rather complex and risky at best. I rather dither when not needed than accidentally disable dither when it would be needed.

I read that post, but because when I read it I still had a false preconception, it didn't clear up my confusion. Now, the post makes much more sense.
For what it's worth, those were questions fueled merely by curiosity, because I've come to greatly appreciate Volume Leveling myself.

[fitbrit]

https://en.wikipedia.org/wiki/Attribution_bias#Hostile_attribution_bias

Thanks for that. It will be useful in the future, no doubt.

[stewart_pk]

Hi, yes I have a 24 bit DAC but I have a USB to SPDIF converter that outputs only 24 bit but the ASIO driver only accepts 32 bit or at least it is able to. So how do you/we know that it only uses 24 of it's 32 bit input?

I guess this is my point, I don't know if it truncates 8 bits from 32 input and then outputs 24 from or it dithers down from 32 bit input to 24 and then outputs it. And if it's actually the latter I'll have to ask my original question again in that is it better to turn dithering off as to prevent double dither? And if the latter is true (32 bit input into the USB to SPDIF ASIO driver is dithered down rather than truncated to 24 bits) then dither or no dither then I believe it to actually be the technically superior option.

Am I at least making sense? If I'm talking rubbish I'd just prefer to know.

[stewart_pk]

have you been following this post? http://yabb.jriver.com/interact/index.php?topic=100095.0

Sort of.

[AndyU]

I'm going to let this go a little longer but close it soon. 

JRiver staff doesn't have time to be discussing all the details of dithering, and I personally think we've reached the very edge of what might possibly affect audio quality.  I don't want our development effort focused on all the things that "might" affect quality.  We need to draw a line and I think we're already a little bit past that line.

Or as we say in Minnesota, we have bigger fish to catch.

Sorry, if I've moved the thread off-topic, and please leave it open long enough at least for Bob Katz to post his measurements. fwiw I think it is excellent that you have responded to his request, which was considered and well argued IMO, and I'm happy to trust his judgement until I'm in a position to use my own.

[Hendrik]

Double dither at 24-bit is not as bad as potentially not dithering at all. Unless you can proof it one way or the other, I would go with the safer option myself.
In a similar situation, many ASIO drivers request 32-bit even when the DAC is only 24-bit, including mine. I can't be sure if they dither at some stage, so I prefer to tell MC to dither to 24-bit, just to be on the safe side.

[stewart_pk]

Thanks for all the replies. I will leave dither on. I may be wrong but I'm going to assume for now that the Thesycon ASIO driver does in fact support true 32 bit data and not just 24 bit data in a 32 bit package (padded). Once again assuming this is the case then I think outputting 32 bit is the technically superior option.

[JimH]

I'll close it now.  Bobkatz, if you need to post measurements, let me know and I'll open it again (briefly).  I'm jimh at jriver.