Selig Leveler Question

This forum is for discussing Rack Extensions. Devs are all welcome to show off their goods.
User avatar
NMN
Posts: 3
Joined: 21 May 2021

Post 05 Jun 2021

I have just purchased this brilliant, time saving RE and I have one question about finding the proper audio 'valley' for a track. Is there some other plugin, RE or display indicator in Reason that might help me to find a good 'valley' value for a track? The Leveler manual says to use the Peak Level display number to find the peak of the quietest word or note. So, for example, on a lead guitar track, I click the Peak Reset button and play notes that look like they are the quietest based on the waveform of the track. This gives me a Peak Level for that note on the Leveler. However, the waveform is not always a good indicator of what the dB level is and so it can sometimes take me a while to search through all of the quiet notes to find the quietest.

I guess I want to try to get a 'Peak Valley Number' along with the 'Peak Number'. I don't know if this would even be technically possible, but wondering if anyone has any ideas for obtaining this value more 'automatically'.

User avatar
integerpoet
Posts: 227
Joined: 30 Dec 2020
Location: East Bay, California

Post 07 Jun 2021

I bet Selig will be along shortly to make a better suggestion than what I'm about to type.

But I recently did an eight-voice acapella project with a piece in which the quietest parts weren't even discrete words and wobbled all over the place from quiet to oh-so-very-quiet. I surveyed each track and got a feel for the numbers and made a rough estimate and listened to the result. Came out more than fine. I feel like the approach taken by the tool is inherently so much better than compression-with-makeup-gain that you will be way ahead even without an absolutely perfect decision.

User avatar
guitfnky
Posts: 3466
Joined: 19 Jan 2015

Post 07 Jun 2021

I think Integer has it right--eyeball it, and estimate. then see how it sounds (start to finish), and make sure nothing jumps out to you as out of place. make tweaks from there if you need to. if it sounds good, it is!

User avatar
NMN
Posts: 3
Joined: 21 May 2021

Post 08 Jun 2021

Thanks for the replies. Finding the proper dB valley manually is still definitely easier than leveling and compressing manually, so still excited about this RE. As I was going through my tracks with the Leveler, it just made me wonder if there was a better way to find the value than what I was using.

User avatar
integerpoet
Posts: 227
Joined: 30 Dec 2020
Location: East Bay, California

Post 08 Jun 2021

NMN wrote:
08 Jun 2021
Finding the proper dB valley manually is still definitely easier than leveling and compressing manually, so still excited about this RE. As I was going through my tracks with the Leveler, it just made me wonder if there was a better way to find the value than what I was using.
I suppose there might be a quicker way if Selig wanted to analyze for signal-vs-noise, somewhat like iZotope RX can, and make a recommendation. But if you're sophisticated enough to care enough to use Leveler, it seems likely your judgment is pretty good.

User avatar
MrFigg
Posts: 6508
Joined: 20 Apr 2018

Post 08 Jun 2021

Way I got it explained is that you’re not actually looking for the quietest sound. You’re looking for the quietest useful sound. You might not want to bring up the volume of a breath for example which may be the softest sound in the recording. Also if you raise all the words to the same volume it’ll probably sound unnatural so use the blend to keep some of the dynamics in there.
丰2ॐ

User avatar
selig
RE Developer
Posts: 9214
Joined: 15 Jan 2015
Location: The NorthWoods, CT, USA

Post 11 Jun 2021

MrFigg wrote:
08 Jun 2021
Way I got it explained is that you’re not actually looking for the quietest sound. You’re looking for the quietest useful sound.
MrFigg nails it - there is technically no quantifiable "quietest sound", so you have to define what is "useful" yourself. I REALLY wanted an easy way to automate this function when initially designing the device, but soon realized there is no technical definition of "quietest sound" like there is with "highest peak" (which is super easy to automatically identify via Peak/Hold metering). In the same way you set a compressor to give you what you think is the right amount of compression but STILL have to listen through the track to be sure, or that you set a threshold for a gate but STILL have to listen through to make sure there are no false triggers, the same applies here. Here's my approach…

The idea, using vocals as an example, is to first be easily able hear what you're doing to the audio with the Leveler, which is best accomplished by increasing the Blend to the max so you can more clearly hear the results. Basically, if the Curve amount is too small, some words will be "dropped" (not raised), and if it is too large you will hear extraneous noises (breaths, bleed, noise) being raised to objectionable levels. It typically only takes one pass to hear any problem areas, then you can simply focus on those areas to fine tune the settings. Then use that same section of your song to adjust Blend to a more natural amount based on the fact that this section is where you're doing the most "leveling" (raising levels).

Short cut: in my experience many vocals fall between 12 and 24 dB of difference between the loudest and softest words, often between 18 and 24 dB (especially when not recorded with compression). So if you want to save time just start with a Curve setting around 18-20 dB or so and check from there. If your voice is the main voice you record, and you aim for a consistent highest peak level for each recording, you'll soon find that you can almost "pre-set" the Leveler because your settings will be very similar from song to song.

Bottom line, learning what a too-high vs too-low Curve settings sounds like is the key here.
Selig Audio, LLC

User avatar
NMN
Posts: 3
Joined: 21 May 2021

Post Yesterday

Thank you for the replies Mr. Selig and MrFigg! I feel better knowing that I'm not missing some additional method to find the valley level. Your explanation makes sense. Should I also interpret from your response that the "Quick Vocal Leveling Setup" section in the manual is a quick way to get a starting point, and then I should work on the ear training to adjust the final curve? I started out thinking more in terms of exact curve values with the Leveler vs. approximate values with a standard compressor. I was thinking the blend would be the most custom setting. When I tried the Multi-Band Combinator for vocals, I started to realize that values may not have to be so exact because I sometimes get three different Peak Levels on each of the three Levelers, but the Curve rotary adjusts all three simultaneously. Thanks again for this amazing RE!

  • Information
  • Who is online

    Users browsing this forum: CommonCrawl [Bot] and 0 guests