# Leedh digital volume?

#### JustJones

##### Senior Member
Forum Donor
I was looking for a network streamer to use with Dutch & Dutch 8c. AES3 out don't need a DAC and was looking at a Lumin U1 mini. I noticed they mentioned this new Leedh digital volume processing. I am not very good at reading through some of the BS and was curious if anyone here has any experience or can explain in layman's terms if this is mainly hype or has promise?

Last edited:

#### mansr

##### Major Contributor
It's all BS. Digital volume control is multiplication. There is no such thing as better multiplication.

OP
J

#### JustJones

##### Senior Member
Forum Donor
Makes sense it was basically subjective opinions on how much better it sounded which tossed my BS switch.

#### thingswelike

##### New Member
Believe it or not, we actually try really hard to keep marketing BS out of the LUMIN material, but it's not always easy lol.
It's not easy to describe Leedh Processing in layman's terms, but I'll try my best:

Let's assume we have these values for a volume level:

-12.0db, 64591
-12.1db, 64468
-12.2db, 64300
-12.3db, 64189
-12.4db, 64045
-12.5db, 63886
-12.6db, 63700

Let's also assume that I want to reduce the volume slightly - say from -12db to -12.5db

Traditional volume control algorithms (including our previous one) aim for volume adjustment accuracy, so it will just take -12db and -12.5db (for 0.5db adjustment). But Leedh processing aims for exact numbers, so in this case we will use -12.2db and -12.6db instead (note the zeros on the end). This means the adjustment is not exactly 0.5db each step, but the values are more "exact".

The rationale is when the value reaches the DAC chips, most DAC will simply truncate those least significant digits, hence there will be data loss using normal approach. While a more "exact" value will have a better chance or retaining the whole value through the DAC process.

So basically, the volume values are represented by "as few bits as possible" to give more headroom for other DAC processes.

We've received overwhelmingly positive reviews from users - so much so, that many are choosing to sell their pre-amps and go direct to power amps.
It's no coincidence that this innovation came after we released our own power amp to be used directly connected to LUMIN streamers.

Last edited:

#### mansr

##### Major Contributor
Well, that there is a great example of bullshit. A digital volume control is super-simple:
1. Multiply each sample by the gain factor.
2. Dither to the resolution of the DAC.
That's it. Sure, depending on the gain factor, some input values will end up with an "exact" output value while others will not. Since music contains samples of pretty much all possible values, you can never get all of them to be exact after the volume control. That is why you must use dither.

This Leedh thing is nothing but a Leech on your wallet.

#### RogerM

##### Member
So by choosing particular scaling factors to be used to set volume level you aim to reduce the +-0.5 lsb error value the DAC hardware converts to analogue ?
I can see this kind of makes sense, but I am skeptical the +-0.5 lsb error from standard volume control can be heard. Especially since the input digital signal must already contain such an error (nature does not work in 16bit values).

#### Berwhale

##### Major Contributor
Forum Donor
Believe it or not, we actually try really hard to keep marketing BS out of the LUMIN material, but it's not always easy lol.
It's not easy to describe Leedh Processing in layman's terms, but I'll try my best:
Welcome to the forum and thanks for taking the time to explain how Leedh volume works on Lumin products.

#### mansr

##### Major Contributor
Details here, I guess

https://www.processing-leedh.com/copie-de-presentation

@mansr that doesn't totally look like BS to me
Looks can be deceiving. That explanation starts from a flawed (or at least contrived) assumption, applies a few individually correct but irrelevant steps of logic, arriving at a nonsensical conclusion. The whole argument hinges on the output having significantly higher resolution than the input. With 24-bit recordings in abundance, this can hardly be considered the norm.

The correct, and trivial, way to do digital volume control is to use an intermediate precision higher than that of the output and apply suitable dither before the final rounding. With a 24-bit DAC, this dither will be well below the noise level of the electronics, so the question of whether or not it is exact lacks any relevance.

#### thingswelike

##### New Member
Thank you Berwhale!

This Leedh thing is nothing but a Leech on your wallet.
Leedh Processing was added to the entire back-catalogue of LUMIN streamers free of charge, with an easy way to toggle it off and on in the App for A/B testing.
There is no requirement to use it and of course there are many many users (such as myself) who have multiple sources connected to a pre-amp and leave the volume attenuation turned off in LUMIN.

#### mansr

##### Major Contributor
Leedh Processing was added to the entire back-catalogue of LUMIN streamers free of charge
One way or another, someone is paying for it. There's no free lunch.

#### Berwhale

##### Major Contributor
Forum Donor
I needed a refresh on dithering and quantization from Monty, here's a link to the relevant part of one of his videos:

#### ElNino

##### Senior Member
This isn't snake oil -- it's actually kind of a cool idea. The guy behind this is Heeb Thierry, who was also the mathematician behind the Anagram ASRC.

A better way for most people to understand it intuitively is that if you start with a 16 bit signal and have 24 bit output, if you limit your volume control to bit shifting, you don't actually need to add dither because you're not introducing additional quantization.

In the AES paper, they still do propose dithering at the last stage to preserve a constant noise floor, so any advantage to the system is a little more subtle than the intuition above (and requires more math), but you get the idea.

Where I think the company's marketing falls down is that it isn't true that most DAC ICs truncate after 24 bits these days, but if S/PDIF is involved somewhere in the chain, then that is true.

Last edited:

#### Mnyb

##### Addicted to Fun and Learning
Forum Donor
Is modern DAC internaly really limited to 24 bit integer internally ? Would they really not work in Floating piont ? And the internal representation in a DS DAC is it not some weird 5 bit oversampled stuff ? Just curius if this simple 16 vs 24 bit analogy actually works ?

#### ElNino

##### Senior Member
Is modern DAC internaly really limited to 24 bit integer internally ? Would they really not work in Floating piont ? And the internal representation in a DS DAC is it not some weird 5 bit oversampled stuff ? Just curius if this simple 16 vs 24 bit analogy actually works ?
No, most modern DACs use a 32 bit representation and have at least a 48 bit accumulator. They don't use floating point because there's no need for it -- they're not doing equalization so they don't need the headroom it can provide.

The analogy I gave is just for intuition to motivate why this approach may be more accurate, but the math is more involved.

#### PierreV

##### Addicted to Fun and Learning
Forum Donor
Where I think the company's marketing falls down is that it isn't true that most DAC ICs truncate after 24 bits these days, but if S/PDIF is involved somewhere in the chain, then that is true.
Yes, to be fair, the marketing always breaks down in terms of audibility, as it does for about 50 to 60% of the DACs Amir has tested here.

#### mansr

##### Major Contributor
This isn't snake oil -- it's actually kind of a cool idea. The guy behind this is Heeb Thierry, who was also the mathematician behind the Anagram ASRC.

A better way for most people to understand it intuitively is that if you start with a 16 bit signal and have 24 bit output, if you limit your volume control to bit shifting, you don't actually need to add dither because you're not introducing additional quantization.

In the AES paper, they still do propose dithering at the last stage to preserve a constant noise floor, so any advantage to the system is a little more subtle than the intuition above (and requires more math), but you get the idea.

Where I think the company's marketing falls down is that it isn't true that most DAC ICs truncate after 24 bits these days, but if S/PDIF is involved somewhere in the chain, then that is true.
Strictly speaking, it does what they say it does. There's just no reason to be doing that in the first place. The basic premise (16-bit input) is frequently not met, so the argument falls apart before we even begin.

S/PDIF is 24-bit, yes. With the inherent noise level of even the best DAC, that's enough that you don't even need dither, though using it doesn't hurt either. The whole thing is an elaborate solution for a nonexistent problem.

#### ElNino

##### Senior Member
Strictly speaking, it does what they say it does. There's just no reason to be doing that in the first place. The basic premise (16-bit input) is frequently not met, so the argument falls apart before we even begin.
The intuitive explanation I gave is not the "basic premise" behind the system. The basic premise is that this does reduce quantization artifacts, at least prior to dither. The AES paper quantifies this, but specifically doesn't claim it is or isn't audible.

OP
J

#### JustJones

##### Senior Member
Forum Donor
I'm not sure what DAC is in the D&D 8c. What I was wanting to do is go straight AES3 from the streamer to the speaker input using a split AES3 y cable with impedance matching. So I don't think it will go over 24/192 anyway. I need a volume control in there somewhere. This looked interesting but I'm not a mathematician or engineer. Some made sense and some read like marketing. I appreciate Lumin chipping in.

#### ElNino

##### Senior Member
I'm not sure what DAC is in the D&D 8c. What I was wanting to do is go straight AES3 from the streamer to the speaker input using a split AES3 y cable with impedance matching. So I don't think it will go over 24/192 anyway. I need a volume control in there somewhere. This looked interesting but I'm not a mathematician or engineer. Some made sense and some read like marketing. I appreciate Lumin chipping in.
Doesn't the D&D DSP have a built-in volume control? That'll be the best place to apply it, because it'll be applied at the full resolution of its DSP pipeline. If it's a hassle to use whatever smartphone app they provide to control the DSP volume, maybe get an iPod Touch just to run the volume control app and leave it on your equipment rack.