Levelling the volume on two vocal tracks in Reaper

[ajay_m]

well... yes, but there may also be an issue where one presenter is further away from the mic and there is noticeably more 'room ambience', which the human brain tends to interpret, understandably, as the presenter being 'further away'. So if the other source is much more intimate there's a disconcerting switch between virtual acoustic spaces, likely to be made worse if the more ambient vocals are boosted to equalise the volume levels.
Ideally in that scenario you want to look at getting the recording ambience consistent between the presenters at source.
It *might* be possible to fiddle around with a very short tail reverb to move one of the presenters to be further away in the perceived acoustic space perhaps, I've never really tried to do this, but having worked with training material I can attest to how disconcerting it is for the listener if the acoustic space switches abruptly with the presenter.

[Matt Houghton]

True. I was assuming a relatively acceptable sound. OTOH, if using computer based remote recording software, while the rooms will sound different, usually the distance of person to mic doesn't vary too much (unless someone's using a headset/lav mic). But if one part's really ambient and the other bone dry it will sound horrible anyway! The answer is to make better recordings, or use de-reverb processing in post. There are great tools for tackling ambience now — DeRoom Pro, Clarity vxDeReverb, Descript etc. These tools are superb (DeRoom Pro is my favourite for this, but the lite version of DeReverb will do a decent job on most dialogue). If you don't want to make both parts bone dry but to reduce the difference, you can even copy the reverb off one part and put (some of) it on the other with like Chameleon and it's pretty convincing. Kinda depends on the aesthetic you're going for and what level of artefacts are acceptable, but it's not hard to get acceptable, fairly dry podcasting results with such tools now. They've come a long way in the last couple of years.

[Drew Stephenson]

I didn’t mention lowering the peaks. You’ve done that yourself in your head.

Er, no. The OP mentioned it in the last sentence of the post you replied to. Hence my clarification.

"In the past, I've been cutting either side of a peak, pulling the fader down on it a little, then glueing the track back together, but as you can imagine, that takes AGES."

[Wonks]

I didn’t mention lowering the peaks. You’ve done that yourself in your head.

Er, no. The OP mentioned it in the last sentence of the post you replied to. Hence my clarification.

"In the past, I've been cutting either side of a peak, pulling the fader down on it a little, then glueing the track back together, but as you can imagine, that takes AGES."

Cross posting I think. I was replying to an earlier post, that post by the OP seems to have appeared before mine, though it either wasn’t there when I was writing/posting or else I didn’t get notification there was another post.

[shufflebeat]

…have you considered using a combination of ears and faders?

I wondered what they were for.

In the past, I've been cutting either side of a peak, pulling the fader down on it a little, then glueing the track back together, but as you can imagine, that takes AGES.

I’ve set up a little custom action (happy to share) which I use to speed up automation which inserts four envelope points on the “volume” envelope (select track, press V) at time selection, resulting in a little handle I can grab and subdue some enthusiastic content.

Select your problem audio
Display envelope (press V)
Click envelope to select
Apply shortcut -creates 4 points
Grab envelope in time selection and attentuate to taste.

[Zukan]

*cough* sidechain comp/expand *cough*

[nightofjoy]

Have you installed the SWS extensions? If you haven't then go to the Reaper site and get hold of them because they add some useful functions to Reaper.

One of those functions is Loudness. You can highlight all the items that you want to match, go to Extensions->Loudness and then click the analyse selected items button. You can then right click on the loudness information and choose Normalize... This will bring up a dialogue box where you can set the level that you want. I would suggest bringing the louder one down to the level of the quieter one so use the lower number in the Integrated column.

While a quick method like this may not be the absolute ideal solution, it will get you very close in just a few seconds. You may also want to experiment with compression if the voices are very dynamic and maybe a bit of limiting if you need to bring the level up but I wouldn't go overboard on processing.

Hi.

How do I get to EXTENSIONS>LOUDNESS...I'm on a mac and Extensions isn't one of my top bar menu fields... thanks.

[nightofjoy]

It's exactly the same effect as normalising, though one is automatic.

I assume you're talking about pulling the fader down. Cutting peaks and pulling them down is more like compression than normalisation.

Back to the OP, Reaper has a setting to automatically apply a cross fade when splitting a track, have you got this enabled? It speeds up that approach hugely.

But if you're really just pulling down peaks then consider something like Gclip if you're on a PC. Really simple and transparent way of bringing down transients.

So where I find a huge peak, I make a cut either side of that peak, then pull the fader down on the isolated peak section, then glue it back into the track.

I know it isn't the correct way of doing it, and when I do it a lot, it makes the audio sound a little choppy, because of course, I've not only lowered the peak, but a fraction of audio either side of the peak, and lowered that too, so when it's back in line with the rest of the track, it causes a very slight jump.

If there was a way of just bringing down the actual tops of any peaks across the whole track, that would be ideal.

Also, links to any video tutorials on this stuff would be really helpful, my brain is starting to feel a little fried!

Thanks.

[Matt Houghton]

If there was a way of just bringing down the actual tops of any peaks across the whole track, that would be ideal.

This is what a peak limiter is for — a processor that brings down any signal that goes over the threshold. A limiter is just a compressor with a very high ratio.

But you still want to get the two parts in the same ball park loudness-wise before you start doing stuff like that. That's where either meters and ears or LUFS normalisation (or, actually, all the above) come into play.

If you find you're knocking more than a few dB off with a limiter, it's a sign the audio is too loud in the first place — just use clip gain (or loudness normalisation) to bring it down a bit more until you're not knocking so much off the peaks.

(I'm assuming this is dialogue here, not sung vocal parts? For the latter there's a whole world of compression and parallel processing to learn that I wouldn't dream of suggesting for dialogue...)

[James Perrett]

How do I get to EXTENSIONS>LOUDNESS...I'm on a mac and Extensions isn't one of my top bar menu fields... thanks.

You need to install the extensions first. I'm not a Mac user but here's a video from a normally reputable source that should show you how to do it.

https://www.youtube.com/watch?v=NDwmqJYJq9s

Hope that helps.

[shufflebeat]

…it makes the audio sound a little choppy, because of course, I've not only lowered the peak, but a fraction of audio either side of the peak, and lowered that too, so when it's back in line with the rest of the track, it causes a very slight jump.

If there was a way of just bringing down the actual tops of any peaks across the whole track, that would be ideal.

To do this in automation just apply three (rather than my previously suggested four) envelope point onto the volume envelope and drag the centre point down where the peak shows on the waveform. You’ll end up with a rapid fade out/in that’s adjustable to taste.

The other answer, as mentioned by others, is any number of compressor/peak limiter units that will do the job automatically, although the result should be checked as one setting isn’t always appropriate for the whole track and adjustments may be necessary, which can also be automated with the envelope system.

[nightofjoy]

OK this has all blown my mind. Far too much information for my poor brain to take in. I'll have to try to sift through it all for the next episode, no time to do anything this time, need to get my episode out.

Quick question, if I use a levelator type thing called Auphonic, at the end of the edit, do I use that on just the dialogue tracks alone, or on the final master with all music and effects?

Thanks.

[Drew Stephenson]

Looking at the brief description on the site, I think you could use it on both the individual tracks and, if necessary, on the masterbus as well - though hopefully it wouldn't need to be doing very much on that.
EDIT
Try it and see but I'd recommend using on a loudness setting on the individual channels (it should then be easy to balance the different channels by ear) and then a true peak setting on the masterbus just to catch anything that has snuck through.

[Matt Houghton]

…it makes the audio sound a little choppy, because of course, I've not only lowered the peak, but a fraction of audio either side of the peak, and lowered that too, so when it's back in line with the rest of the track, it causes a very slight jump.

If there was a way of just bringing down the actual tops of any peaks across the whole track, that would be ideal.

To do this in automation just apply three (rather than my previously suggested four) envelope point onto the volume envelope and drag the centre point down where the peak shows on the waveform. You’ll end up with a rapid fade out/in that’s adjustable to taste.

The other answer, as mentioned by others, is any number of compressor/peak limiter units that will do the job automatically, although the result should be checked as one setting isn’t always appropriate for the whole track and adjustments may be necessary, which can also be automated with the envelope system.

The other other answer is editing. If you're going to go through an do it manually (as with automation) then I'd suggest just using edits and autofades instead of automation. Create a custom action/mouse modifier to select with a swipe/click drag on the waveform then use a mouse modifier to run an action to select the clip under the mouse cursor and reduce by x number of dB. (And run it again if you need to). Assuming you have autofades enabled, it's a similar result to automating, but (a) probably quicker and (b) allows you to change the levels before any processing (though you could also automate the pre gain instead of the fader of course).

It's not the same thing, but I wrote a workshop a while back on podcast editing in Reaper for which I created a few useful custom actions — you should be able to figure out from those how to roll your own...

[shufflebeat]

…it makes the audio sound a little choppy, because of course, I've not only lowered the peak, but a fraction of audio either side of the peak, and lowered that too, so when it's back in line with the rest of the track, it causes a very slight jump.

If there was a way of just bringing down the actual tops of any peaks across the whole track, that would be ideal.

To do this in automation just apply three (rather than my previously suggested four) envelope point onto the volume envelope and drag the centre point down where the peak shows on the waveform. You’ll end up with a rapid fade out/in that’s adjustable to taste.

The other answer, as mentioned by others, is any number of compressor/peak limiter units that will do the job automatically, although the result should be checked as one setting isn’t always appropriate for the whole track and adjustments may be necessary, which can also be automated with the envelope system.

The other other answer is editing. If you're going to go through an do it manually (as with automation) then I'd suggest just using edits and autofades instead of automation. Create a custom action/mouse modifier to select with a swipe/click drag on the waveform then use a mouse modifier to run an action to select the clip under the mouse cursor and reduce by x number of dB. (And run it again if you need to). Assuming you have autofades enabled, it's a similar result to automating, but (a) probably quicker and (b) allows you to change the levels before any processing (though you could also automate the pre gain instead of the fader of course).

It's not the same thing, but I wrote a workshop a while back on podcast editing in Reaper for which I created a few useful custom actions — you should be able to figure out from those how to roll your own...

I might be misunderstanding the process but doesn’t this end up with the instant gain drop/recover that the OP was trying to avoid rather than a fade out/in?

[simonmix]

I think you need compression and level normalization. It would be good to be able to extract the two voices on separate channels.

[shufflebeat]

It's not the same thing, but I wrote a workshop a while back on podcast editing in Reaper for which I created a few useful custom actions — you should be able to figure out from those how to roll your own...

That's a neat little custom action, by the way. I'll be incorporating a version of that for entirely different operations, including adding FX/processing to certain items and not others.

[Hugh Robjohns]

I can't help think this is all getting way more complicated and time-consuming than it needs to be!

1. Record conversation to separate tracks.
2. Adjust levels by ear to match and balance, leaving adequate headroom.
3. Edit out the unwanted crap.
4. Tweak the levels of any isolated sections which are overly loud or quiet (eg, riotous laughter, mumbles and asides, etc)
5. Insert gentle compressor / limiter on each track or the final mix to give modest dynamic range control and elimination of any big transient peaks.
6. Export mix.
7. Job done. Get a beer....

The only time-consuming and fiddly part should be No.3 The rest is trivial and simple.

[Matt Houghton]

I can't help think this is all getting way more complicated and time-consuming than it needs to be!

1. Record conversation to separate tracks.
2. Adjust levels by ear to match and balance, leaving adequate headroom.
3. Edit out the unwanted crap.
4. Tweak the levels of any isolated sections which are overly loud or quiet (eg, riotous laughter, mumbles and asides, etc)
5. Insert gentle compressor / limiter on each track or the final mix to give modest dynamic range control and elimination of any big transient peaks.
6. Export mix.
7. Job done. Get a beer....

The only time-consuming and fiddly part should be No.3 The rest is trivial and simple.

Yep, that's how I go about it! Was just offering advice because for some reason it seemed OP not satisfied with it!

[Matt Houghton]

I might be misunderstanding the process but doesn’t this end up with the instant gain drop/recover that the OP was trying to avoid rather than a fade out/in?

Depends how you do the edit. If you're not moving parts, just separating into different regions so you can apply clip gain, then you can manage the transition with cross fades.

But as Hugh says, this all feels a lot like hard work compared with the usual way of doing this stuff

[Wonks]

I'd be wary of just looking at waveforms and meters and just close your eyes and listen. Voices do go up and down when talking. Level the vocals too much and it can start to sound unnatural, boring and hard to listen to.

If you both learn to set up and record at similar levels, then you may find that some gentle overall compression (so things don't get too loud or too quiet) are all you need.

[shufflebeat]

I can't help think this is all getting way more complicated and time-consuming than it needs to be!

1. Record conversation to separate tracks.
2. Adjust levels by ear to match and balance, leaving adequate headroom.
3. Edit out the unwanted crap.
4. Tweak the levels of any isolated sections which are overly loud or quiet (eg, riotous laughter, mumbles and asides, etc)
5. Insert gentle compressor / limiter on each track or the final mix to give modest dynamic range control and elimination of any big transient peaks.
6. Export mix.
7. Job done. Get a beer....

The only time-consuming and fiddly part should be No.3 The rest is trivial and simple.

Yeah, it’s number 4 we’re concentrating on. The rest is pretty straightforward.

The “insert three envelope points and drag the middle one down” is neither time consuming or complicated.

Tea, thanks.