Issues with Selig Leveler

[Last Alternative]

Let me start by saying I LOVE this thing and use it all the time. However, there are a couple of issues I have with it. Keep in mind I have the auto and look ahead settings on. Also, it is the first thing in my signal chain on vocals and I record straight from mic to Focusrite 2i4 and I record @ 88.2.

1) Clearly yellow means the value the first knob should be set at -> whatever the highest number it registers is. Blue is for what it's worth, the silence. Green is the number the horizontal fader should be -> again, whatever the highest number is. My issue is that it is rare for that green number to pop up. Sometimes it doesn't at all, even though my vocals are clearly jumping around in volume. This makes it extremely hard to know where to set it. I go by ear but then it's breathy here and there, etc. It's just easier when it's clear and the green number holds.

2) This is more of a recommendation than an issue. I feel there should be some way to hold the highest number of each color. Like a yellow dot and a green dot you click and it lights up to show which one you are holding, then an off hold switch or something. Sometimes the numbers and colors jump around so fast you can't see where you should set to.

I feel it could use an update. ...or I'm using it wrong.

[pjeudy]

This is more of a recommendation than an issue

I was just thinking that! Maybe the thread title should also say..suggestion ? I thought there was a bug with Leveler..I also used often

[Gaja]

Yes that would be awesome!
As it is now you have to find the quietest word manually and then measure only that word. If that part could be automated, thenentire device could set up itself. Just throw it in, play the song once and in the end it sets target and curve values all by itself and all you need to do is adjust time, dry/wet and output level.

[selig]

Let me start by saying I LOVE this thing and use it all the time. However, there are a couple of issues I have with it. Keep in mind I have the auto and look ahead settings on. Also, it is the first thing in my signal chain on vocals and I record straight from mic to Focusrite 2i4 and I record @ 88.2.

1) Clearly yellow means the value the first knob should be set at -> whatever the highest number it registers is. Blue is for what it's worth, the silence. Green is the number the horizontal fader should be -> again, whatever the highest number is. My issue is that it is rare for that green number to pop up. Sometimes it doesn't at all, even though my vocals are clearly jumping around in volume. This makes it extremely hard to know where to set it. I go by ear but then it's breathy here and there, etc. It's just easier when it's clear and the green number holds.

2) This is more of a recommendation than an issue. I feel there should be some way to hold the highest number of each color. Like a yellow dot and a green dot you click and it lights up to show which one you are holding, then an off hold switch or something. Sometimes the numbers and colors jump around so fast you can't see where you should set to.

I feel it could use an update. ...or I'm using it wrong.

First, ignore the colors - they mean nothing like what you've described, and I hope it's not something I've said…

If I've somehow given you a totally wrong impression of how to use the Leveler, go back and read the back panel again for clarity on the simple setup process.

Short answer: INGORE the color, and go by the LEVEL of the incoming audio. Set the Target to the highest level you want to include in the upwards compression (typically but not always the highest peak level of the signal), and set the Curve to the size/width of the area below the target you wish to increase. Done.

You don't REALLY have to measure for the softest word to find the "proper" Curve setting. If I'm listening to the Leveler and I hear breaths as loud as the vocals, I know the Curve setting is too high. If I hear words that are "missed" by the Leveler (too soft), I know the Curve is too low. In the extremely rare case I hear both, I automate the Curve setting!

As for the colors, if it's rare for a green number to appear, your levels are hotter than they need to be (but this my recommendation, not anything hard and fast). As a guide to setting levels, green indicates "good", yellow indicates "warning". Blue just indicates a low level signal, not noise. These colors are used as in Selig Gain too, fwiw.

[selig]

Yes that would be awesome! As it is now you have to find the quietest word manually and then measure only that word. If that part could be automated, thenentire device could set up itself. Just throw it in, play the song once and in the end it sets target and curve values all by itself and all you need to do is adjust time, dry/wet and output level.

That's a big IF. Folks keep suggesting I add this feature like it's at all possible to do automatically. But consider this: how would the software know to "include" a soft word but ignore a soft breath? They look identical to the meter… ;(

This is one of those things the human ear can do instantly, but the algorithm that would have to be written to accurate discern between a sound that is "valuable" and one that is "not" would be immense and far outside of the scope of my understanding of audio! Not saying it's not eventually possible, but it's not been done to date.

The other issue is that any "auto" settings cannot be saved with the device. That means you would have to play the entire song once through each and every time you loaded it before it could set itself up correctly. I'm guessing that would not be a desirable "feature" for most?  

[Tincture]

EDIT: I wrote this while Selig was posting obviously, so kind of redundant

It took me a while to understand this device fully I feel. Def. in my top 3 fave REs.

I think you've got it Jason but I think about it in a different way maybe?? I think I have this right but someone please correct me if not! ...

Whatever the number levels out as (I have it hold the loudest), say -10dB, is the loudest part of the vocal... so if you want leveler to try and match that level for the range you set then yes, set this as your target... but I often like to leave my target a dB or two lower than this (e.g. in this case -12dB) to leave a bit more dynamics.

The horizonal Gain is 'how far below (in dB) the target value the device will "reach" to bring up the gain' ... this is the bit that it took me a while to get my head around. So say you want every tiny lipsmack or tongue click to be heard go-ahead and use 24dB of gain (think 24dB quieter than the target being attempted to be brought up to the target level). But if you just want a more dynamically even performance just take the gain up slowly until the quietest part sounds loud enough. I never bother doing it by the numbers.

I hope that didn't come across as condescending, it's just that maybe writing out my experiences might help others avoid some of the head-scratching I had for a while?

[Exowildebeest]

Yes that would be awesome! As it is now you have to find the quietest word manually and then measure only that word. If that part could be automated, thenentire device could set up itself. Just throw it in, play the song once and in the end it sets target and curve values all by itself and all you need to do is adjust time, dry/wet and output level.

This is one of those things the human ear can do instantly, but the algorithm that would have to be written to accurate discern between a sound that is "valuable" and one that is "not" would be immense and far outside of the scope of my understanding of audio! Not saying it's not eventually possible, but it's not been done to date.

It probably has been done in the high-tech world... With flight-data/cockpit voice recorder analysis to discern the breathing of the co-pilot from the banging on the door 

(too soon?  )

[selig]

EDIT: I wrote this while Selig was posting obviously, so kind of redundant

It took me a while to understand this device fully I feel. Def. in my top 3 fave REs.

I think you've got it Jason but I think about it in a different way maybe?? I think I have this right but someone please correct me if not! ...

Whatever the number levels out as (I have it hold the loudest), say -10dB, is the loudest part of the vocal... so if you want leveler to try and match that level for the range you set then yes, set this as your target... but I often like to leave my target a dB or two lower than this (e.g. in this case -12dB) to leave a bit more dynamics.

The horizonal Gain is 'how far below (in dB) the target value the device will "reach" to bring up the gain' ... this is the bit that it took me a while to get my head around. So say you want every tiny lipsmack or tongue click to be heard go-ahead and use 24dB of gain (think 24dB quieter than the target being attempted to be brought up to the target level). But if you just want a more dynamically even performance just take the gain up slowly until the quietest part sounds loud enough. I never bother doing it by the numbers.

I hope that didn't come across as condescending, it's just that maybe writing out my experiences might help others avoid some of the head-scratching I had for a while?

What you write makes sense, AND it points out that there is obviously room for improvement in the interface to the Leveler so that the two control's function are more clear. It's indeed interesting a device with basically two controls can be so confusing, but I'm in agreement that this is the case to some degree! 

So if anyone has suggestions as to how these controls could be made more clear, I'm open to hearing them!

[stratatonic]

[>So if anyone has suggestions as to how these controls could be made more clear, I'm open to hearing them!

Maybe an option when the Leveller is calculating the Peak, the Target knob can follow accordingly (Peak showed -7dBFS - Target knob moved to that level as well)?

[Gaja]

Yes that would be awesome! As it is now you have to find the quietest word manually and then measure only that word. If that part could be automated, thenentire device could set up itself. Just throw it in, play the song once and in the end it sets target and curve values all by itself and all you need to do is adjust time, dry/wet and output level.

That's a big IF. Folks keep suggesting I add this feature like it's at all possible to do automatically. But consider this: how would the software know to "include" a soft word but ignore a soft breath? They look identical to the meter… ;(

This is one of those things the human ear can do instantly, but the algorithm that would have to be written to accurate discern between a sound that is "valuable" and one that is "not" would be immense and far outside of the scope of my understanding of audio! Not saying it's not eventually possible, but it's not been done to date.

The other issue is that any "auto" settings cannot be saved with the device. That means you would have to play the entire song once through

each and every time you loaded it

before it could set itself up correctly. I'm guessing that would

not

be a desirable "feature" for most?  

Yeah I thought something like that as well. Still if it would work like that it would be pretty awesome

[selig]

[>So if anyone has suggestions as to how these controls could be made more clear, I'm open to hearing them!

Maybe an option when the Leveller is calculating the Peak, the Target knob can follow accordingly (Peak showed -7dBFS - Target knob moved to that level as well)?

You can't move a control from the DSP code - only via automation or Combinator. 

All of these ideas were floated three years ago during development. Not that it bothers me to be asked and answer these at all, just letting you know they didn't just slip under the radar…  

[Tincture]

IMO I don't think anything needs relabelling or explaining ... for me it was just a matter of understanding the concept of upwards compression correctly... which for me mainly meant realising that Gain is the number of dB below the Target that you want Leveler to attempt to bring up to the target, rather than somekind of make-up gain. Hopefully that doesn't confuse anyone!

[guitfnky]

That's a big IF. Folks keep suggesting I add this feature like it's at all possible to do automatically. But consider this: how would the software know to "include" a soft word but ignore a soft breath? They look identical to the meter… ;(

This is one of those things the human ear can do instantly, but the algorithm that would have to be written to accurate discern between a sound that is "valuable" and one that is "not" would be immense and far outside of the scope of my understanding of audio! Not saying it's not eventually possible, but it's not been done to date.

would you be able to have it function based on whether or not the sound has a detectable pitch? obviously that would mean reworking the RE to include some sort of pitch detection (if that's even possible), and it wouldn't work for whispers, really, but just thinking out loud (with my fingers).  or alternatively, I wonder if there's some sort of CV you could use to have another existing RE or stock unit tell the Leveler whether it's receiving pitch.  again, not sure if that even exists, really.

that said, I wouldn't need such a thing.  Leveler is amazing as-is.  

[Gaja]

Leveler is amazing as-is. [smile]

That is true for me as well.