Boosting volume after the fact

spantini

COO of me, inc.
I know how to normalise tracks during and after mixing, but I had a mix I rendered to mp3 which came out at a low volume. Not too low, it was reasonable, but lower than most people's mixes. When playing on Win 11's Media Player in my laptop, I had to set the volume at 99 to match other's mp3s which could be played at a volume setting of 70.

I loaded the stereo render mp3 back into a Reaper track and tried to normalise it but it wouldn't normalise.

So instead of going to a limiter, I used an app to boost the dBs by X-amount (3dB in this case), and that corrected the situation. Now it's volume matches that of other's mp3s - volume setting of 70 is plenty loud.

This got me to wondering what those apps use to raise the dBs. A limiter? If so, what settings would they use? The limiters I've used in Reaper do the job well but are easily overdone.
 
I think it's simply done by maths, adding a specific number to each sample. I think that's also why it only takes one sample glitch - the ones you see when you look closely at waveforms - to prevent a track being normalised. In my examples, often normalising reduces a quiet track even further and almost always there is are a few isolated spikes, that often cannot be heard, or maybe a peculiar DC onset for a few samples that raises the maximum. I notice that these are perhaps introduced into the music by a sample glitching in one track, and it gets right through to the mix down. Cubase's master fader overload lights in the channel strip are triggered by these and sometimes I just have no idea why a quiet track suddenly has an overload warning. Sound Forge is quite good at looking for these and zapping them. Looking by eye at a magnified waveform takes forever and sometimes you still miss them. It always makes me smile that you can take a few samples and simply remove or flatten them and you cannot hear what they actually did to the music.I suppose it's like old DAT error correction where it gave you a display of how many samples were being error corrected, and until it got over a thousand a second, you could not hear what was being fixed.
 
I'm not rendering the MP3. The MP3 is a render which has had no tracks normalised pre-rendering. I dragged the MP3 to a new Reaper track and tried to normalise that but it wouldn't. Just wanted to see if it would work and how it sounded.
 
Throw it in Audacity and see if there is a particular peak that hits 0dB. You should also have seen it in Reaper during rendering. If so, that's why it won't normalize. Take the file that you used the app boost and look at the same peak. Did the program just clip the top off, or did it do some type of limiting to keep an approximation of the wave peak, but to drop the level. I think with Audacity, you can actually allow it to clip peaks, so if there are only a couple of peaks that are limiting the level, you can just clip them off. They may not even be audible.

Normalization simply looks at the file and finds the highest peak level, and then adjusts everything to make the highest peak to be 0.0dB (if you normalize to 0). Its been years since I read the mathematics involved in adjusting an audio signal (can't remember if it adds or multiplies the number by a factor).

The nice thing about a digital compressor/limiter vs analog is that it can look ahead and do all the calculations before it ever applies the adjustment.
 
I'm not looking at the render at the moment, but I'm sure there is at least one peak. Sometimes I get it down to two or three, but there always seems to be one at the start - probably my MT Power Drums.

Normalization simply looks at the file and finds the highest peak level, and then adjusts everything to make the highest peak to be 0.0dB (if you normalize to 0). Its been years since I read the mathematics involved in adjusting an audio signal (can't remember if it adds or multiplies the number by a factor).
I was reading about this also. I haven't analyzed any of this, I'm just lightly playing around with it.

In the future I'll be normalizing tracks before rendering. I'll begin with normalizing all tracks and see where that goes. This is fun.. :p
 
The other part that you might play with is make up gain. With a compressor, you can end up with a track that is slightly lower in level than the original. Make Up Gain will bring the level back up.

I usually use the compressor on my MTPDK snare and kick. I also drop the level of them relative to the toms and high hat.
 
Yeah, same here with using MTPDK's snare and kick compressors. I believe Reaper's compressors - some at least - let you lock the makeup gain to the threshold. I haven't done this yet, just been working each separately by ear.

I'm working on building a drum kit in ReaSamplOmatic5000 with a bunch of free samples I've scavenged. These won't need as much taming and shouldn't give me those peak problems. If any do, the 5000 can easily eliminate them.
 
The other part that you might play with is make up gain. With a compressor, you can end up with a track that is slightly lower in level than the original. Make Up Gain will bring the level back up.
When its left reduced. no make up. the wav becomes thicker. this is important.

Also look at the ceiling used. Lower the ceiling, the limiter wont pump it so radically at level.
 
I know how to normalise tracks during and after mixing, but I had a mix I rendered to mp3 which came out at a low volume. Not too low, it was reasonable, but lower than most people's mixes. When playing on Win 11's Media Player in my laptop, I had to set the volume at 99 to match other's mp3s which could be played at a volume setting of 70.

I loaded the stereo render mp3 back into a Reaper track and tried to normalise it but it wouldn't normalise.

So instead of going to a limiter, I used an app to boost the dBs by X-amount (3dB in this case), and that corrected the situation. Now it's volume matches that of other's mp3s - volume setting of 70 is plenty loud.

This got me to wondering what those apps use to raise the dBs. A limiter? If so, what settings would they use? The limiters I've used in Reaper do the job well but are easily overdone.
Generally, you'd set the output level ("loudness") of your mix in the mastering step using a combination of compression and limiting, along with whatever you might feel is needed.

"Normalizing" is usually just a matter of adjusting gain to some maximum peak value, and I'm not sure what "it wouldn't normalize" means in your post, but it's not really the same as setting the loudness.

Ozone (iZotope) has a set of tools, and the "standard" version is half price ($124.50) right now. If budget allows, it might e something to consider, as it would let you target a specific loudness, or adjust it to your own target(s).
 
I went back to that MP3 I was trying to normalize. All this discussion has caused me to take a closer look at what actually happened. When I said nothing happened (it didn't normalize), I was basing that on what I observed as the waveform appeared not to have budged at all. Just tried it again and took a closer look.. it did change, but not noticeably - looked like nothing happened.

Up next to the item's Vol knob it says in micro-print. . . -.35 dB !!! So it did normalize:facepalm::p

I tried a different MP3 and it reduced the waveforms quite noticeably. I then opened an old project and marquee selected several tracks and normalized - they all reduced. Another project and the tracks actually increased quite a bit. Hmmm...

I did all that with the normalizing at default 0dB settings. I'm beginning to get somewhere and learning something in the process.
 
Getting back to my OP - I was curious to know how 3rd party apps boost the dBs of one of my MP3 files - the MP3 being the "after the fact" part of the thread title. I'd like to know if they do this with a limiter.. or what? IDK....
 
?? My son invariably sends me MP3s of his classical guitar pieces at -30dB FS and I just whack the fader up on the mixer in Samplitude and "save as".

Dave.
 
If some sites, like YouTube, strive to set the streamed audio output within a range, say -14dB LUFS (+/-), it depends, of course, on what the source audio's LUFS is, but they likely have a process, similar to a plugin, that simply massages the compression and gain and shoves it through a limiter. I suspect the processing, if smart enough, may do very little other than gain, especially if lower the LUFS, or even if raising it and the file's current peaks will not exceed -1.0dB FS (e.g.) if only gain is needed. But, just guessing.

More recent versions of iZotope's RX has a "Loudness Control" (only in Standard or Advanced - no longer a standalone) that will allow you to set it and process an entire audio file. There are other tools and plugins, or you can simply put a LUFS measuring plugin, like Youlean's, at the end of your [mastering] FX and adjust gain/EQ/pan/compression/etc./limiting until you hit your target, while keeping the true peak below 0dBFS. For all I know, (Reaper already has this stuff.)

 
Thanks, Keith. I'd guess streaming sites are as you describe. Reaper does have the LUFS plugins - I have been going over the videos. One way I've been using to get a level in Reaper is to use the render window's meters and peaks/LUFS indicators to hit -14. I do several dry runs, nudging the master +/- until it hits -14, then I render.
 
Back
Top