LUMIN Music - Leedh Processing

Don’t you have “DSP volume” available in device controls ? Right-click on device icon / device setup

I don’t think that’s available for RAAT [edit: unless running stock Roon Bridge]. Directly connected / local device and RAAT device have different volume mechanisms in Roon.

All my RAAT devices have DSP volume control

Not on my Lumin

But it’s there on my tablet.

Yeah, like I said, it depends on the device situation.
Roon has to deal with a complicated set of devices and connection modes.
And sometimes those are flawed: a USB device is supposed to declare its capabilities and the streaming gadget it’s connected to is supposed to report it back, but sometimes there are flaws.

1 Like

Yep, no DSP volume leveling selectable with Lumin (just Device).

The iPad allows for DSP and Device volume control.

Roon labels all these variants as “DSP Volume” and qualifies as “High Quality” only, probably as it’s not floating (I learn something new here everyday).

In my test I was playing back vinyl that had been ripped at 24/382 with an RME ADI-2 Pro ADC. I first played them in my normal mode (Lumin [A1] - TRL Dude Pre - TRL Samson amps - Green Mountain Audio C-3 HX). Then I re-cabled to remove the TRL Dude Pre. All the cabling was Silent Source. I need to do more testing but my initial reaction was that there was more clarity and detail and also more body to the instuments and more dynamics with no apparent trade-offs so far (other than being limited to one source input.).

You’re not wrong because the integrated volume control in a classical preamp/poweramp or integrated amp configuration, is an integral part of the preamplifier part. When you keep all these preamplifier parts in the chain you keep the possible bottlenecks intacts and the possible virtues of the leedh volume control maybe can’t be heard even if you turn your amplifier volume controle to the max.

So for me the use of leedh volume control goes with a reflection about the signal path in each audio system, to try and decide where are the bottlenecks, maybe if a direct path between dac and power amplifier can be the transparency choice (if dac and power amplifier have a good relationship… impedance, tonal balance, right gain, no need for an other amplification part to add something to listening pleasure but that is damaged by a classical volume control to attenuate this one more amplification part).

For now I use the leedh processing on a U1mini with a Ayon Skylla 2 dac and 813 single end tube power amplifiers with Marten Duke2 speakers in a home office. For me the leedh processing give better results in this configuration than the classical digital volume control of the Lumin, and than the integrated volume control of the Skylla 2 dac. Some years before, I used a Ayon Crossfire integrated amplifier with a motorised ladder volume control and connecting the Skylla2 to the Crossfire by the direct input, bypassing the ladder and using the Skylla volume control, gave far better results in transparency (so take care of the preamplifiers and integrated, the bottleneck is often here behind the pleasure of a colored or limited item). Today the leedh volume control give a bit more relaxed presentation with better tone accuracy, better resolution on fine details, increased naturalness in voices, acoustical instruments that seems less compressed, less planed at one end instead of being gradually attenuated over the entire strip with respect of each levels of dynamics.

Is this a high fidelity volume control at the end? Honestly I don’t know. But all other volume controls are not too, so the deal is always to make a choice between compromise volume controls, and I think the leedh’s one has something special that may give more natural feelings and open windows on the musical work, if the signal path allows it in the audio system.

Next try for me, the U1mini coupled with my Tambaqui dac, luxman M800’s and Egglestonworks Andra’s, with or without my Coincident Statement preamp.

1 Like

ok, so… giving Leedh a try in my system

  • Metrum Adagio’s volume knob set at 3/4 (Metrum states this is the way to use the Adagio as DAC only. don’t ask me why :no_mouth: )
  • Lumin U1 Mini as Roon endpoint
  • Roon’s volume slider

well… Leedh ON/OFF difference is very obvious: transparency, body, detail, “air”… :slight_smile:
switching among the two is just a click away and volume level is preserved
pity you can’t do it on-the-fly during playback (playback stops an play position goes back to track’s beginning)

difference between Leedh and Adagio’s own volume control is harder to determine: more passages are needed, takes more time and volume level is not retained
but later I’ll try comparing these too :wink:

4 Likes

ok: tried Leedh Processing vs. Metrum Adagio’s own VC (which is nor digital nor analogue: works by “variating DAC’s reference voltage” so… not touching the signal itself)

well… being sure is really hard as a few passages (… and time) are needed to swap among them and, even more, because spl is not retained so one needs to re-adjust the volume

I tend to prefer (slightly) my Adagio’s VC. I find it slightly more transparent, full bodied and maybe has also a little more detail
though, as I said, tiny differences in spl… two controls being summed anyway (when using Leedh)… expectation bias… poor memory :stuck_out_tongue:

in any case… Leedh’s doing a terrific job, that’s for sure! :slight_smile:

3 Likes

I’ve looked over their materials. I understand the idea. It produces numbers that are pretty to look at, and fits well into a marketing message about being “mathematically lossless”. I don’t think it does harm.

The marketing message is black+white…mathematically lossy is “bad”, their product eradicates it. Previously digital volume control was problematic, now it is the equal of analog domain volume control thanks to this innovation. Plus, volume control in general prevents hearing loss and tinnitus :stuck_out_tongue: .

Poking fun aside, I have some more technical thoughts.

One problem with their line of reasoning is that it assumes “mathematically lossy” processes like truncation are inherently problematic. We have many good reasons to alter audio signals irreversibly, and lots of mathematically lossy stuff happened in recording/mixing/mastering, and more happens in the DAC chip post volume control. This doesn’t mean that information loss is “good”, it just means that to discuss this topic honestly, you have to consider the magnitude of information loss, the places that it happens in the processing chain, the nature of the information loss at each point, and how it plays out end-to-end, from the recording studio all the way through to your ears.

These marketing materials focus on one attribute of the process–mathematical reversibility, treat it as a black+white benefit, and then focus only on one step in the chain in isolation, and then from this narrow vantage, make an argument that their approach addresses the past problems of digital volume control because it doesn’t leave you with any non-zero bits to truncate away.

If I look at floating point volume adjustments through the same narrow lens as these marketing materials, floating point comes out even better: floating point multiplication within the ranges used in audio processing is mathematically lossless without requiring imprecise volume adjustments. No special techniques required.

(Of course they aren’t arguing against floating point processing or even considering it in their arguments–they are implicitly comparing to an integer-math volume control, because they are more common in hardware products).

I don’t use FLACs because they are “lossless”. I use FLACs because after ripping to FLAC, I have captured 100% of the information on the CD so I can stick it in a box in the attic and never have to worry about ripping it again. I know that the original mastering process was not lossless anyways, so I am under no illusions and I happily accept that even the best DACs perform mathematically lossy operations that lose significantly more information than the truncations discussed by Leedh because it is just part of how audio is reproduced. It is not so black and white.

One final point–the main reason why digital volume control can be a disadvantage is not that it sometimes involves mathematically lossy processes. It is because of where it sits in the processing chain. It necessarily consumes some of the overall signal-to-noise ratio of the end-to-end system. The analog components are a much stricter bottleneck for SNR, and a quieter digital signal put through the same analog stuff operating at the same amplification level is going to to have a smaller effective dynamic range, whereas an analog amplifier with a volume adjustment can be designed to have roughly the same SNR across a range of volume levels.

Those effects aren’t in “quibbling about bits and truncation” territory anymore. If you had 100dB of usable SNR to start with, and you chop off 20dB of it digitally with any technique, the quality difference will be in “audible to a layman” territory regardless of how it’s done.

In most systems there will be some space for digital adjustments without doing meaningful damage, but the core implication of these marketing materials is that the longstanding “problem” with digital volume controls is addressed by this innovation, and I’m not sure how something like this could go that far. I can say that I won’t be pursuing it for my own listening, based on their published information.

I have been collecting user feedback of our implementation of Leedh Processing Volume on Lumin product line. I don’t want to spam this forum with a lot of those posts. Interested readers may check out https://www.audioshark.org/lumin-169/lumin-leedh-ing-way-18376.html#post304204

I’ll just post some detailed review(s) here.

1 Like

This is an evaluation of Lumin X1 Leedh Processing Volume by Wesley Miaw of Neko Audio LLC:

I strongly prefer the new firmware with Leedh Processing Volume. There is less blurriness / quicker decay and less masking of sounds across the entire audible spectrum. The overall presentation is more balanced (more midrange presence compared to the old firmware)—I suspect because there is less distortion which was creating more noticeable additional energy in the bass and treble. With better detail retrieval and clarity, it is much easier to understand vocals and to hear minute details that were previously hidden.

That being said, I can imagine some people who do not listen to those things wondering where the bass went, having become used to or to expect more bass presence at the expense of accuracy and precision. And maybe also feeling like there’s less “air” because the treble has been cleaned up.

Sounds are more realistic with the new firmware.

My process was to listen to a minute or two from each track of my test playlist, sequentially, using the old firmware 12. I did this twice, to get acclimated to the old firmware sound and ensure it wasn’t just listening to a track a second time that was changing what I heard.

I then switched to the new firmware with Leedh Processing Volume and listened to each track again, sequentially, for a minute or two. Then switched back to the old firmware, and finally again to the new firmware as a final confirmation.

Lumin X1 → Lumin AMP → Magico M2. The volume was set at 55/100 the entire time. The main listening position is about 9’ from the speakers, arranged so my ears are on-axis with the tweeters. The top of the seat back is below my ears to minimize immediate reflections. Room is heavily treated so reflections are minimized. Rear surface is a blackout curtain about 5’ away from the main listening position.

Here is my test playlist. I use these tracks because they highlight / focus on specific things, and can reveal weaknesses.

Weight of the World / 壊レタ世界ノ歌
河野万里奈 NieR:Automata Original Soundtrack

神様のいうとおり
いしわたり淳治 & 砂原良徳 + やくしまるえつこ 神様のいうとおり

Cups (Pitch Perfect’s “When I’m Gone”) (Pop Version)
Anna Kendrick Ultimate Pitch Perfect

ラビリンス (Album Mix)
MONDO GROSSO 何度でも新しく生まれる

1.618
BT This Binary Universe

The Demon God
Joe Hisaishi Princess Mononoke

Madder
Groove Armada Lovebox

Teardrop
Massive Attack Collected

Highly Strung
Orianthi Believe

goodblue
school food punishment amp-reflection

Force [Live]
Superfly Force

Prokofiev: Piano Concerto #2 In G Minor, Op. 16 - 3. Intermezzo: Allegro Moderato
Yuja Wang; Gustavo Dudamel: Simón Bolívar Symphony Orchestra Of Venezuela Rachmaninov: Piano Concerto #3; Prokofiev: Piano Concerto #2

기억 (Original Mix)
윤하 Someday

意味/無/ジュニーク・ニコール
SQUARE ENIX NieR:Automata Original Soundtrack

片翼の天使
植松伸夫 Distant Worlds: music from FINAL FANTASY

Magic Hour
Ahn Trio Lullaby For My Favourite Insomniac

Leedh processing may result in better sound quality, but who wants to give up a physical remote control for an software based, Ipad experience that is less responsive? My fingers hurt too much as it is from constantly touching an Ipad and computer keyboard. Shifting between other apps on a tablet to tweak the volume and back for a small improvement in SQ is not appealing. I wonder if Lumin has considered creating a physical remote for their music players? Now that Leedh has been released, it might help sell a few more Lumin Amp Stereo Ampliferers.

I don’t see a remote in the future, who needs another remote? The selling point for me was not just the sound of the T2 but also the incredible futuristic design, so yeah not a big fan of physical remotes and I don’t think it will add any additional value.

1 Like

I’d like to add that Leedh Processing Volume at Lumin volume 55 is roughly -22.6dB, while non-Leedh volume is -22.5dB, i.e. Leedh Processing lowers volume than standard by 0.1dB in this particular comparison.

It is generally accepted that a little higher volume would let people perceive something to sound better. In this particular example, Leedh Processing Volume -22.6dB is actually at a theoretical disadvantage compared to the ES9038PRO volume -22.5dB. When a trained listener prefers a digital volume algorithm at a lower volume than another digital volume algorithm at a higher volume, I believe it validates the superiority of the former.

I expect objectivists will come out and argue this is just an expectation bias. But no, what I originally asked Wesley Miaw of Neko Audio LLC to do is to compare the SQ of two firmwares, without telling him what the differences are. He did not know he was testing Leedh Processing Volume before we announced it in public. His report above originally referred to the two firmwares as “beta firmware” and “release firmware” - for readability I changed the “beta firmware” to “new firmware with Leedh Processing Volume” and “release firmware” to “firmware 12”.

In another case, I sent a beta tester a firmware that decreased the volume to comply with Leedh Processing Volume, but the beta tester said it sounded louder. I guess this is caused by lowering of noise floor due to no longer suffering losses of least significant non-zero bits by the DAC. This was the moment I know we are doing something right.

@AndersVinberg
FYI. The web site added a white paper and a presentation done at the last AES of beginning of June by SUPSI (Swiss university).

Thank you Peter for all the detailed explanations on firmware comparison. Out of curiorisity : in your “standard” digital volume control implementation (not LP) are you doing prior to ESS9038 or using the embedded volume control of the ESS ?
Also have you compared SQ at digital volume -20dB or -40dB ? These are values were LP and “traditional” volume control should be identical.
Congratulations to Lumin anyway for continuously improving and a reasonably priced high end line of products.

For analog outputs in X1 and T2, we are using ES9038PRO and ES9028PRO built-in volume control if Leedh Processing Volume is disabled. (By the way, a Lumin X1 user carefully evaluated the X1 standard volume control against his preamp that features relay-controlled attenuators, and found the X1 to be superior, before Leedh Processing Volume was added.)

I cannot confirm that. These values look conforming only in decimal.

Thanks for the upgrade!
Thought the T2 already sounded very good, but it made it even better.

2 Likes