Let’s agree to disagree.
I treat my room, I use Dirac Live digital room corrections, I have ample dynamic range in my system, yes - in the low end just as well (to cater my favorite bass heavy music genres) and still - I use “loudness” when listening at low levels like in the small hours of the night and find it much more enjoyable.
I don’t use it at reference levels, which happens automatically since this is how it works when calibrated properly. This is how it is designed.
What I don’t understand is why do you come to a feature request thread to bash a feature request that is legitimate, straightforward to implement (in the limited form suggested), is implemented by top gear and competitor software and is scientifically backed?
As a reference here are two excerpts from the JRiver feature announcement back then:
Loudness provides a more natural sounding volume control based on the frequency response characteristics of human hearing.
When the Loudness feature is enabled in Media Center, it uses (roughly) the ISO 226:2003 curves as the volume is decreased to provide a more linear sounding volume…
…
…
…
At the reference level (100% by default), no correction is applied.
I employ JRiver’s loudness control. Set it so it has not effect at reference level (i.e. 83 dB SPL), but as the volume is turned down, the amount of loudness control kicks in to boost the lows and highs, so it still sounds balanced, even at low listening levels. Works pretty well for me.
OK, fine. It appears we have somehow different experience and opinions not only on loudness, which I personally view as an implementation of human hearing deficiencies not needed specifically on high level reproduction system, but also on what electronic corrections can do and cannot do. I see them as excellent first-order corrections, but not able unfortunately to fully overcome limitations and issues of speakers, room and their interaction. I have been using these corrections and I still use them, so do not please misread a word of caution as an attempt to bash or prohibit.
At this point let me also praise Roon DSP implémentation whose insertion in the digital path does not introduce any perceptible transparency loss, at least with the high accuracy USB converter I am using. That alone is a remarkable point and an enabler for using such corrections (with caution) in a high fidelity reproduction system. It is a bonus in favor of Roon and a demonstration of the solidity of their implementation.
When people post about the sound quality of Roon they should distinguish between verifiable things like DSP, leveling, room correction, loudness–things that should always be optional, never defaults–versus categories like boutique cables and magic black boxes.
It used to be common to see loudness controls on audio gear. But there was rarely any provision to achieve level calibration (the Quad system one poster described must have been one of them). I’ve always used separate pre and power amplifiers, but it’s rare for a power amp to include a level pad. For that reason, loudness always sounded to extreme for my ears, and I avoided it. My preference has always been to listen at realistic levels whenever feasible. That includes clavichord music, for example, which often has to compete with ambient noise.
If Roon came out with loudness I might give it a try, but it would be complicated by the fact that I normally don’t use Roon to control the volume, and still listen to other sources like silver and black disks. I’ve used JRiver for a long time for ripping and tagging only, but rarely for PB, and didn’t know it had loudness.
Agreed. A Roon implementation of loudness seems difficult in all cases where, to benefit from full numerical resolution, the volume control is made outside Roon. With other software it is the same issue and as you said, analog loudness was more common in the past but full implementation requiring absolute dB calibration was not following. Then it was abandoned on strict high fidelity implementations, much alike tone controls, to a degree replaced by non-neutral interconnects and other tweaks to fine-tune the tonal balance of systems.
Brian mentioned that (external volume) as a holdup for implementation at some point… By it shouldn’t hold anything, not all features are supported in all modes and that’s understandable.
I think it should be pointed out that this approach only works when you know the intended replay level of any given album/track. Which is never (rarely) given, or known. Basically every album, or track, would need to come with a sensitivity (Pa/“virtual amplitude”) [virtual amplitude] = 1
But even if you did know it, playing it back at any other pressure level would lead to inaccuracies, as loudness is not time invariant. Loudness is quite tricky for non-stationary signals. Please refer to last years revision of ISO 532 (ISO 532-1:2017)
Furthermore head-related transfer functions would also need to be considered.
Loudness implementation, to use the term “scientific” with some accountability, should be applied with an absolute calibration in decibels. Unless one has no external control volume, the risk of shifting the curves is real, be it subjectively pleasing or sounding artificial. Hence it would appear rational to propose it in association with some calibration.
Actually if made self-adaptative and dynamic with some calibrated mike, loudness could be used in all configurations… by those who find it appropriate on their systems, which is not a claim for universal benefits.
Again, many audio people, including very experienced ones, do not feel any need for loudness compensation and rather avoid it despite (or because) they listen across a large® dynamic range and spectrum. Like Mitch and Leporello pointed out, not only reproduction systems and rooms can differ (enormously) but also our listening capabilities, discernibility and taste. Just like deep neural networks can give very different estimation results, based on different training sets…
All this takes us back to the overstated title of this thread, which is probably why I felt some reaction was needed. The term “scientific” should not be used as a marketing argument, and the thread would have been less controversial if labelled more like:
“On my system, at low level, I prefer Jriver with dynamic loudness correction, to Roon, without”,
which describes your argument correctly, right ? Or was it just to make the buzz ?
What I said about Mitch was about the variability of systems, not about using or not loudness (or a loudness proxy in the absence of level calibration).
That me and others pursuing the idea of an unbiased sound reproduction don’t feel the need for loudness, does not meant be against other people using such features.
I gave some feedback about other digital correction systems and problematics like room and system response corrections which appear sometimes confused with loudness corrections. I also thought it would be the right place to provide some comments on how active loudness correction could (if this is in Roon’s team wishlist) be implemented with some rigor, so perhaps now a possible “Roon” loudness feature request is better documented. At least if it uses the same libraries and data flow as he DSP features of Roon, we should have a transparent implementation of it, which is certainly most desirable. Likewise, one could also think of inserting a clean digital dynamic expander/compressor into Roon.
But I prefer the minimalistic approach - better tune a system so that it can play any good recording at its best with a bare minimum amount of digital tweaks. Deactivating all should be a very minor setback. I know signal processing and no amount of it cannot fully compensate non-linear electrodynamic transducer distorsions and complex mechanical and acoustic interactions. Better reduce all this in the physical domain : there are still many good years for excellence manufacturers in audio.
And like just said Lars, careful with claims in audio domain.
My claims are in the logical domain, not audio domain.
Could have made the exact same claim when Roon didn’t have DSP with PEQ/Convolution/Whatever audio related feature - and it would be correct. Whoever has the feature CAN sound better. Can meaning not always, sometimes. In some conditions. To some people. But now it has DSP. In the future maybe Loudness with calibration/tilt will join the party.
It could still be misleading. It suggests that it is possible to configure Jriver in such a way, that it (objectively) sounds better than not configuring it, or compared to Roon. You have also added “it’s scientific”.
I maintain that the term “scientific” should never be used lightly, like a marketing argument.
This is true in general but particularly in audio.
It is very difficult to have a thorough scientific approach in audio:
because it combines complex electronic, electrodynamic and acoustic topics + bioacoustics
because our ear is extraordinarily sensitive to non-stationary phenomenons
because of the difficulty to design measurements in full correlation with subjective appreciations
because of the variability of the equipment and audio configurations for listening sessions
because our brains really “interprets” the audio signals in very elaborate and variable ways
I stop there, more could be added.
I am sometimes concerned by the perceptible lack of rigor and experience, frequently readable between the lines, in audio forums and Internet discussions. Also part of it can be driven by commercial arguments, sometimes by people who are both juge and party. I would not like to see that develop in that community, for the following reason.
I have no commercial interest in Roon other than being a personal client but am extremely pleased by what Roon and associated tools like HQplayer are currently offering on digital audio. On this basis, I would personally not like Roon threads to be lead in directions that would not incline to preserving the purity of signal reconstruction that it enables. It takes some years and some rigor to achieve the kind of result achieved by Roon.
One interest of these threads, is to allow ideas and experiences to be shared, with not only pros but also cons, from discussions, ideas can be synthesized and bared down to their essential concept - or sometimes appear as not as universal or relevant as initially thought. That’s not only normal, that is a sign of vitality and health. Ideally this type of forum becomes a place where products can be improved, novelty can be proposed, old ideas can be revisited, and people with less experience or insights can learn from people with more experience or insights.
I fail to see what is stopping the OP from designing a few Parametric Equalizer curves, based on the loudness reseearch mentioned, for some wanted listening levels?
On the other hand, a interactive loudness option would be nice! You all remember Yamaha and their variable loudness adjustment? It was meant to be used as a compliment to the volume control, first set requested listening level, then add loudness if wanted. A bit like seasoning your dinner at the table.
