Jriver Loudness implementation, where are the down sides?

[daemon d]

I noticed an article somewhere about the Jriver 18 loudness control, which gave a quick and very basic run down on how they say it works and why it's so wonderful.

 

I haven't tried it, but we all know the Fletcher Munson curves from 1933 and the ISO 226 equal loudness contours (which are non-same), so in theory a non-linear loudness equalisation makes sense; especially to me where I get in trouble for pushing the volume just a bit much in order to hit the sweet spot, from my usually patient wife.

 

However, the Jriver implementation must apply cuts in the digital domain; does this decimate actual data, even though they say their 64bit digital volume is pure and good?

 

Does it introduce differential phase shifts (bad) or does it phase correct at the expense of processor overhead and latency (less bad)?

 

What sort of audible bad things if any, is it likely to introduce?  Interested in your opinions.

[Nada]

I liked the way it sounded.

 

I cant measure phase changes but my CPU usage didnt change much. Phase changes are meant to blur imaging. I didint notice that.

 

Yes digital equalisers will change some of the encoded volume levels. So what. Most volume levels wont be full level anyway If some are attenuated a few more dB Im not concerned.

[Newman]

Some SNA discussion here, and here. Not specifically answering all your questions, but.

[daemon d]

So I tried it out for a few hours this afternoon: nasty.

 

Upper to mid bass is boosted, but lower bass gets left behind giving a sort of subtle mid-fi bloom and making speakers that were full range suddenly sound like they need a sub.  Treble seemed similar with upper treble being left behind. Imaging contracted strongly in depth and width with the sound stage trapped between the speakers rather than being larger than the room.

 

It did give the sense that mid bass authority was preserved at lower volume, but everything sounded substantially, not subtly, better with it turned off at all volumes. Noticeable audible bad things were definitely present.

 

Oh well, worth a try.

[Nada]

As usual a few minutes listening sorts it out. I must admit I only tried it once myself....

[daemon d]

True true Nada, but not the audiophile forum way old chap: You've got to talk about it; argue about it; agonise over it and then, when you finally listen to it, you have to disbelieve your senses on the word of someone else who has higher magic levels in their ears

[qwerter]

It sounds fine on some recordings. But why would you ask before you try it for yourself? Don't be afraid to press buttons... it is you who will ether like or dislike it at the end.

[daemon d]

There's a couple of reasons I might ask before trying it myself:  One is that there are a few people on the forum who come up with amazingly erudite snippets of fact and theory, which I often enjoy and learn from when I can spot them amongst the crowd.

 

Two is it whiles away the time until I can turn on my stereo.

 

Three, and by far the most important, is that it's the done thing  If seven people with post counts above 1000 had told me it's wonderful, I would be forced to assume my experience of it being vile were either psychosomatic, or that the magic levels in my ears had dropped below critical levels, and thus listen to it no matter how much it hurt.

 

No, really; I was hoping someone new about the data decimation on the 64bit volume control, and the phase thing, since I know it's possible to phase correct for EQ in the digital domain and I was hoping this was so, but thought if the phase mischief was linear and small I might not hear any effect myself, having low magic levels in my ears and all that.

 

Hmm, that's more than a couple = two, isn't it; maybe my magic levels are rising again...

[qwerter]

There's a couple of reasons I might ask before trying it myself:  One is that there are a few people on the forum who come up with amazingly erudite snippets of fact and theory, which I often enjoy and learn from when I can spot them amongst the crowd.

 

Two is it whiles away the time until I can turn on my stereo.

 

Three, and by far the most important, is that it's the done thing  If seven people with post counts above 1000 had told me it's wonderful, I would be forced to assume my experience of it being vile were either psychosomatic, or that the magic levels in my ears had dropped below critical levels, and thus listen to it no matter how much it hurt.

 

No, really; I was hoping someone new about the data decimation on the 64bit volume control, and the phase thing, since I know it's possible to phase correct for EQ in the digital domain and I was hoping this was so, but thought if the phase mischief was linear and small I might not hear any effect myself, having low magic levels in my ears and all that.

 

Hmm, that's more than a couple = two, isn't it; maybe my magic levels are rising again...

Don't take me wrong, did not want to be rude or patronizing...

[hochopeper]

I haven't looked at jriver at all, certainly not looked at their loudness control. Personally I wouldn't use a loudness control for normal listening, though I might use it if I was playing PS3 games or listening to music late at night while my wife was asleep.

 

Well implemented digital volume control easily beats the best analogue volume control though some might make a valid argument that they prefer the sound of the additional distortion of the analogue control. DSP based crossovers can be done better than the best line level analogue or passive crossovers. (again some might suggest that they prefer the distortion of the other types but the measurements are clear, the method that keeps the cleanest signal relative to the input is when the EQ is implemented in DSP). So in terms of EQ I do believe that the loudness control can be done very well indeed, not sure if jriver have done the best possible, do you have a link to details on their implementation to save me searching for info? Regardless, I don't think its something that I'd personally want to use for normal listening though, because it is an arbitrary EQ from the input signal to suit loudness rather than to suit your system/room.

 

In other matters, I never consider post count a sign of experience or knowledge either

[daemon d]

I took no offence Qwerter, I'm just naturally dry with a sarcastic streak (wit of fools and all that). Sorry if I gave you the impression that I took your comment as other than the good natured conversation you intended old boy. If I was actually offended, I'd probably just call you names or something

 

Hocho;  I've only recently begun to take any notice of all the mysteries of digital sound reproduction (which has a lot of room for neophyte confusion as some of the utter tosh people spout on forums indicates), but I was under the impression that digital volume control involved discarding bits from the top down, and when one got to low volume levels, the empty head room bits were all gone and discarded bits included data. My understanding of the 64 bit volume in Jriver is that this gives allot more empty bits to discard, but that one must eventually come to discarding data bits in the end. This is often used, or possibly misused, as the reason digital volume control might be evil or at least not sufficiently audiophile to have real street cred'.

 

Have I got something wrong here? I'm more than willing to let you inform me my friend.

 

I agree re EQ. From what I understand, digital EQ would be the only method where phase could be completely corrected, if it were implemented this way (as it is in things like Acourate, but obviously isn't in Jriver loudness, as evidenced by the strong imaging destruction, which Jriver introduce here  http://yabb.jriver.com/interact/index.php?topic=76608.0  ).

 

Also, I wasn't aware that a good analogue volume control induced any distortion that was realistically audible.

[firedog]

I used it for a while when I was in a situation of only listening at low volumes. I think it helped in that situation - I was enable to enjoy the music at lower volumes without the feeling that it needed to be turned up that I had without it.

 

I do think it reduced the clarity of the sound - but a very small amount. Not something you'd notice in most situations.

[davewantsmoore]

Hocho;  I've only recently begun to take any notice of all the mysteries of digital sound reproduction (which has a lot of room for neophyte confusion as some of the utter tosh people spout on forums indicates), but I was under the impression that digital volume control involved discarding bits from the top down, and when one got to low volume levels, the empty head room bits were all gone and discarded bits included data. My understanding of the 64 bit volume in Jriver is that this gives allot more empty bits to discard, but that one must eventually come to discarding data bits in the end. This is often used, or possibly misused, as the reason digital volume control might be evil or at least not sufficiently audiophile to have real street cred'.

 

Have I got something wrong here? I'm more than willing to let you inform me my friend.

 

I agree re EQ. From what I understand, digital EQ would be the only method where phase could be completely corrected, if it were implemented this way (as it is in things like Acourate, but obviously isn't in Jriver loudness, as evidenced by the strong imaging destruction, which Jriver introduce here  http://yabb.jriver.com/interact/index.php?topic=76608.0  ).

 

 

If you are hearing something from the jriver volume it is not because of "decimating".   64bit volume is one hundred million percent audibly transparent.

 

That is audibly transparent when compared to the same signal, running through the same DSP, just without the volume manipulation.     Comparing it to anything else, is not a fair comparison as there are too many variables.

 

 

 

As for the actual EQ (and linear phase) of the jriver dynamic volume.....  I'm not sure.

 

 

 

Have you considered (I'm sure you have) .... that rather than "nasties" caused by dynamic volume .... that it is just the EQ curve which is causing the effects ?!?    ie.  The EQ curve is being implemented without "fault" ... so the results are correct, but you just don't like them.

[daemon d]

Sure I've considered that Dave. That falls into the sweeping category of nasties.

 

So from a purely theoretical perspective, discarding bits to reduce volume doesn't end up discarding data bits at low level?

[davewantsmoore]

That falls into the sweeping category of nasties.

 

I don't really agree.   Perhaps the EQ curve is not gelling well with your speakers, or room... or your mind.

 

What I'm trying to say is let's keep "intended effects" .... separate from "unintended effects".   Even if you think the "eq curve is nasty" .... then let's call that an "intended effect"     (it's okay that you don't like the intended effect).

 

 

 

So from a purely theoretical perspective, discarding bits to reduce volume doesn't end up discarding data bits at low level?

 

This is not how digital volume works .... given adequate precision  (which 64bit definitely is) ...  So, essentially the answer is "yes".

 

 

Reducing the volume raises the SNR.   If you raise the SNR enough, then you will overrun the signal that the lowest bit was representing.    In practise, if you are using sufficiently more precision in the DSP than the signal, this doesn't happen.

 

64bit precision is mind-blowingly bigger than 24bits.       You must attenuate by 240 dB before you begin to erode the SNR of 24bit audio.... and you must attenuate 246 dB before you "drop the lowest bit".

[davewantsmoore]

Had a quick look on the JR site, and couldn't see anything about the EQ used (ie. linear phase or not) .... but I really wouldn't suspect that is what you're hearing.

 

 

One thing I did notice though was the "reference level setting" ... Did you set that somewhere appropriate?   So the corrections were being applied at the right volume level  (ie.   above a certain level there would be no correction)  ?!

 

I can imagine how setting this too high, would result in excessive correction which might sound bad.

[cada]

I cannot comment on the digital volume being better or worse than good analogue volume etc...but its been a long time since I used a preamp and drive my power amp directly with J river internal volume through an external DAC.

 

I have used internal volume & loudness function leaving the reference on 100% and it does not sound right -  too boomy and inflated in the bass it certainly helps to set the reference level lower say around 50% but I still think it sounds better without it. Without it on I feel I am getting the low levels I need even at very low volumes at night I can still here all the low frequencies without loss.

 

There was a post on this awhile back which I responded the same but it died.  I often wondered if others experience the same.  I have never posted on the J river site and nobody seems to comment on whether they are using this feature or not?

 

If you have small speakers with a bass issue or a room that has a lack of bass, perhaps it would be nice to have.

 

With Audio every circuit, cable, room, speakers add some dimension to the sound so whether digital volume is better than analogue volume I believe cannot be answered.

[daemon d]

Yeah set that Dave. Experimented with it at a variety of settings. EQ can't possibly be linear phase, the change in sound stage is evidence enough of that.

 

Hi Cada, I don't know if digital is better or worse either mate, and I suspect that any difference between good implementations is going to be in the sonically insignificant category, but I agree with your findings on the loudness EQ function.

[davewantsmoore]

EQ can't possibly be linear phase, the change in sound stage is evidence enough of that

 

I do not understand how (the same) EQ applied to both channels, could affect "sound stage".

Not saying you don't hear a difference .... just wondering how sure you are about the cause.

(We may just have terminology issue too of course?!)

 

 

It is very unlikely to be linear phase at an educated guess..... as that would introduce delay (undesirable in a video player).

[Newman]

I do not understand how (the same) EQ applied to both channels, could affect "sound stage".

 

Maybe by accentuating or attenuating frequencies where certain instruments or vocals lie, they can be pushed back or brought forward in the mix?

[davewantsmoore]

Maybe by accentuating or attenuating frequencies where certain instruments or vocals lie, they can be pushed back or brought forward in the mix?

Yes, perhaps it could make them more or less perceived. When I think "sound stage", I think width, height... stuff which is contained in the "stereo" effect. So we're probably just into terminology here.