Hi,

I work on iZotope Rx7. I was wondering if there's a plugin that could automatically equalize the loudness levels of different tracks so that they all sound about same to the human ear. This is for making compilations. I do this manually and it takes lot's of time and effort. If a software could do this for me, heck I'll be relived.

Thank you reading.

 

HoRNet VU meter is excellent auto-gainer, practically free, and you get a nice VU meter, too.

He's also got some other plugins with auto-gain. Like this one. Seems even better for the task because it also sports LUFS metering. Check them all out here.

I think something LUFS based would work better for you, like "TheNormalizer" plugin, but I've been using VU meter for a very long time, and only occasionally have to adjust tracks manually.

 

Last edited: Jan 28, 2019

There's a perceptual loudness function in RX built in. Called Loudness, open it and set to BS.1770-2/3 loudness standard. You may set the target LUFS and peak level, and use RX batch mode to process whole folders of source files. It will scan the waveforms and write perceptually adjusted files out.

Warning, if there are overshooting peaks, they will be limited automatically. Fine as long as it doesn't happen to your files often. With already-mastered compilation tracks and loudness target of -14 LUFS or so, it won't happen at all.

This won't even out the loudness of an actual track, what it does is find the right volume settings for each (simple onetime gain adejustment) at which all your tracks sound like they're the target loudness you set. The track's actual dynamics are left untouched.

 

Last edited: Jan 28, 2019

Okay thank you for the input guys. I'm gonna try both of your suggestions to check which one works better for me.

@ Satai: How much is -24 LUFS in terms of dB ? My target volume varies from 98 dB to 95 dB depending upon the case to prevent clipping. I apologize if I sound confusing, I just wish you could tell me the best settings to have the perceived loudness within the ranges of 95 dB to 98 dB

Thank you again SineWave and Satai for taking time and giving me the valuable input. Highly appreciated.

 

Last edited: Jan 28, 2019

thanks for this, just went on a hornet shopping spree

 

@ceo54 "How much is -24 LUFS in terms of dB?" It's -24dB, which is calculated from average audio strength over a certain period of time. In a VU meter that certain period is 300ms. LUFS calculates it from an even longer period of time, but I cant remember the precise value. K-system calculates it over 600ms periods. What is important to understand is that none of these will show you a peak value, but an average audio strength value.

 

You can play a -14 LUFS file at 95dB or 100dB or even 120dB no problem. Doesn't matter at all to it. The LUFS is something like "what's the average perceptual loudness of the track most of the time?". So it will vary mainly based on the types of sounds in the track and how compressed the producer liked their dynamics to be. If they compressed a lot, then you'll get LUFS readings approaching -9 or -8, because most of the time the track is banging loud. If they were careful and didn't compress too much and it's a delicate ambient track, the LUFS might be around a cool -24.

LU stands for "loudness unit"

FS stands for "full scale", which in digital is anything up to 0db (the max value, where you start clipping waves).

Physical speakers can take this "full scale" digital signal which is kinda abstract, and output it at any actual loudness you want. BUT a -8 LUFS track will still subjectively sound a ton louder at the same volume setting on a speaker, than a -24 LUFS track. Hope it made sense, you're on the right track

 

Don't mix up dBFS and dBSPL, please. Btw, you don't listen to music at 96dBSPL most of the time, do you?

 

LUFSm (momentary) is measured with 400ms period.

LUFSi (integrated) measures the 'whole prgram', could be a track, a spot, an interview. But I couldn't find a 'realtime' period up to now for this one (no idea if it's, for instance, sample-accurate).

LUFS means relative to Full Scale, so it differs with different TP (True Peaks), but it's meant be for TP of -1(dBFS).

 

the topic reads "loudness" and i would be a fancypants and point at the volume fader and say "there you go!" but i can see its a different question, and for that, i dont have a easy answer/way.

use your ears and a eq

 

Audio Vtimain Structure also works, but Hornet auto-gainer is very cheap.

 

Okay so I tried the Loudness module in Rx7 it works but now I hve couple of issues with it. First, it's not precise. I still had to alter the true peak for some tracks manually using a plugin. Second, it first transcodes the audio to PCM.

I have another ancient tool that works directly on the AAC files without transcoding them. The only problem being it doesn't take the avrage loudness in consideration, just adjusts the true peak with the values provided so even though the true peak of the tracks in compilation is always the same but tracks sound differently.

Does anyone know of any other way ? Preferably that works on the AAC files without transcoding them and takes the average loudness of the track and adjust it accordingly.

Thank you for the help. This is an awesome community.

 

Update: I got myself Hornet Autogain but for the life of me I couldn't figure out how do I make it read the value from the one track and apply it on the other track(s) I guess I need a sidechain to accomplish that but there's no sidechain in Rx7

If I give it some static value to work with it works fine, I just have to determine the values of the source.

Thank you everyone for the help.

 

It almost will never be precise to get every track loudness equal to each other via any automation program. The best it can do is to get at the near close ballpark or slightly off. There is where your ear come in to tune each of them so u can get them as close as possible as you desire.

Because here is the thing, doesn't matter if it LUFS, dbRMS, dbFS, dbSPL, dbVU etc, it just a measurement, some calculate the average and some marking by the peak. But none can tell how much loudness each song need because every song has various dynamics, some song has a flat dynamic, and some have a wide dynamic range depending on the parts in the music and also genre specific (ie; slow song, heavy song, bassy music, unplugged).

That's also is the reasons why when it come to remastered an album for other use, let say a compilation or greatest hits, or maybe to do a playlist for live event or broadcast which a variety of mixture song comes in, usually we remastered it again with a minor and mild compression to balance out each track so it has loudness by measurement quite not far from each other. But still, we cannot overdo it, so it won't kill the dynamic quality, at the end we use our hearing to judge the equal loudness of each track especially only on the chorus or busiest part of the song. This is when the measurement does not really matter anymore if it off by 3 LUFS or less.

So to avoid clipping or digital distortion, we take the loudest track as a benchmark and lower its target as a nominal level to compensate for other track, and balance the other by using the new nominal target. If there is still headroom to go for every track after balance, we equally bring it up. Because there is always a track that seems like sounds low but the meter measures it loud, vice versa. And some track can have an over peak 1, 2 or more in a certain part of a song, but it is unusual for a properly mixed and mastered track. Many factors need to take as a consideration, a different time era is one of them. Spotify or perhaps other streaming platform is a good example of bad gain and limiter automation.

You still can use the automated program, its good and it is handy, faster. But don't do too much track at a time, try to organize it. Make by batch or a group, calculate or measure them first and where all the track has quite in the same 6db range, make it one batch, balance it and save. And do the same for the others. It is easier to balance by batch rather than pull everything in one go.

And of coz, your monitoring also needs to be adequate, don't monitor it too loud, it could cloud your judgement.

Sorry if my post, not much of a help to suggest any good software, I'm also still doing it an old school and manual way. But I think all the others already have suggested you great software to get start and job done. I believe what's left is just to organize the workflow and get your talent as an engineer to get involved for the finishing.

All the best to you brother.

 

Last edited: Feb 11, 2019

Nowt finer. Fits over the shaft.

 

That's a lot's of experience packed into that post. Thank you Jeff for taking time and elaborating it to me. Well, the Loudness module is the closest thing that Satai guided me to, I find useful. As you said it's not precise but it leaves little to do compared to doing it manually. I will also keep experimenting with Hornet's Autogainer as SineWave recommended and see if I can find a magical formula that works.

Thank you Jeff. I couldn't have understood that much in a decade that you told me in a single post. And a big thank you to all the people who posted in the thread.

You people rock.

 

Nugen LM Correct. It's a standalone program, not a plugin, but it's great for what you want to do.