Page 1 of 2

Normalisation and compression with two separate voice tracks

Permanent link to this postPosted: Mon Sep 18, 2017 8:44 am
Posted by Thirith
If this has been discussed elsewhere, my apologies; I did a quick search, but I don't know the relevant terminology yet.

I've started a podcast with a friend (you can find the first episode at https://adamnfinecup.com/ - iTunes review pending). As you might expect, I've checked out tutorials and videos on how best to edit the audio and improve the sound; one thing I'm not entirely clear on, though, is how best to do all of this when I've got myself and my co-podcaster recorded separately.

Is it advisable to do some basic maintenance edits first on the two tracks (I'm thinking e.g. of ensuring that there's no clipping) but then first edit the two tracks together, make the necessary snips and tweaks, mix the two tracks and then normalise, compress and equalise the resulting mixed track? Or do I do all of the steps with each of the individual tracks and then edit them together? Or is yet another procedure advisable?

Thanks in advance for your helpful tips!

Re: Normalisation and compression with two separate voice tr

Permanent link to this postPosted: Mon Sep 18, 2017 4:10 pm
Posted by kozikowski
By what method did you coordinate your show? Call each other on the phone, Skype? It's not recommended to record the Skype call. You can do that if you absolutely have to, but it usually doesn't come out very well.
how best to do all of this when I've got myself and my co-podcaster recorded separately.

It's not a contest to see how many effects you can apply. The fewer the better. If your podcast takes off, you'll need to remember to do your effects dance each time and that gets tired in a hurry. Separate tracks is very highly recommended because that's the only way you can make the general quality of the two voices match. That's a very big deal for pleasant listening.

After you mess with each individual track, you can Tracks > Mix and Render to a new Track. That will give you a separate mixed track which lets you apply corrections to both at once. MUTE and SOLO as needed.

When you get a show you like, Select that one corrected track by clicking just above MUTE. File > Exported Selected. It is recommended you Export a WAV (Microsoft) protection copy of the show before you produce the post. I assume MP3. You can't re-edit an MP3 without compression damage and you can't easily make a different compression copy. You can from the WAV. Keep the WAVs of the two original performances. If they're MP3, stop that. Do all original production in WAV.

I need to listen to the sample to see how free-wheeling your presentation is.

Did you find that the two durations were different when you overlayed the two tracks? That's normal. No two recorders ever come out perfect until you get into the Hollywood equipment. You can correct that with little or no damage with Effect > Change Speed. I think that's the one that lets you simply type in the durations you want and it takes care of the rest. Once you do that the first time, it should stay constant from now on—until one of you changes recorders.

So how did you do the show?

Koz

Re: Normalisation and compression with two separate voice tr

Permanent link to this postPosted: Mon Sep 18, 2017 4:19 pm
Posted by Thirith
Quick reply, which I hope will give you a better idea: we talk via Skype but each person records separately, using Audacity and saving the resulting sound file as a WAV.

The first episode suffers from audio glitches on my co-host's side; we've already resolved this, but yeah, the first episode worse in terms of recording quality than I would've wanted. For ep 2, the individual recordings are better thanks to better equipment (new mics, quieter computer).

Re: Normalisation and compression with two separate voice tr

Permanent link to this postPosted: Mon Sep 18, 2017 4:20 pm
Posted by kozikowski
Stop me anywhere if I pass something you miss. I'm going at full steam until I get a sense of how you're working. I sometimes use abbreviations that lose people.

Are you doing all your work in Mono (one blue wave)? Unless there's a valid reason for stereo, mono is a really good idea. My podcast test has stereo music and mono voice. I could be talked into mono everything.

You are Exporting WAV copies of everything against the time Audacity goes into the mud unexpectedly and takes your show with it. There is no: "we need to shoot it again." Just open up your protection copies and keep going.

Koz

Re: Normalisation and compression with two separate voice tr

Permanent link to this postPosted: Mon Sep 18, 2017 4:28 pm
Posted by kozikowski
Dueling posts.
saving the resulting sound file as a WAV.

Oh, so you're way further along than just beginner. Yes, the one downside of this process is needing each actor to be good at recording. There was a recent poster who wanted to know how to save one of the four sound tracks in his show.

Eek!

Can you post two twenty-second sound clips here, one from each voice? Mono WAV, scroll down from a forum text window > Upload Attachment. I'm interested in the latest product from each location. Don't process it. At all.

I wrote a process for audiobooks and it may work for you. They have really tight requirements for sound quality.

As we go.

Koz

Re: Normalisation and compression with two separate voice tr

Permanent link to this postPosted: Mon Sep 18, 2017 4:33 pm
Posted by Thirith
Thanks, I'll do that, though I'll probably only get around to it tomorrow.

Re: Normalisation and compression with two separate voice tr

Permanent link to this postPosted: Mon Sep 18, 2017 4:40 pm
Posted by kozikowski
If you successfully post enough times, forum moderation drops away and posting speed picks up. We're still all volunteers, but moderation is an extra step until we're sure you're human.

You are human, right?

Koz

Re: Normalisation and compression with two separate voice tr

Permanent link to this postPosted: Tue Sep 19, 2017 4:46 am
Posted by Thirith
... well, human-ish.

Anyway, here are the two samples (WAV, mono, zipped so they're under 2MB). They both include a bit of (supposed) silence, but they're mostly talk.

Re: Normalisation and compression with two separate voice tr

Permanent link to this postPosted: Tue Sep 19, 2017 4:47 pm
Posted by kozikowski
I'll listen when I get back home.

There are shortform compressors that might be handy. Chris's Compressor was designed so Chris could listen to opera in the car. It evens out volume variations without seeming to do anything.

More later.

Koz

Re: Normalisation and compression with two separate voice tr

Permanent link to this postPosted: Tue Sep 19, 2017 6:23 pm
Posted by kozikowski
Both of your clips have "rumble" or low pitch tones that aren't useful. It's not always environment (heavy trucks driving by). It's not unusual for some home microphones to make rumble because it's expensive to remove and nobody can hear it. But it can screw up post production filtering. The corrections take into account sound nobody can hear.

Steve wrote a filter to get rid of it without seriously affecting your voice.

LF Rolloff is called Low Rolloff now, but it's the same tool.

viewtopic.php?p=318684#p318684

Ignore SetRMS. That's for a different job.

Matt has very well behaved track and I would have no trouble making him into an AudioBook except he sounds like he's recording in a kitchen or empty room. That's significant because we can't take echoes out of a sound track. So you and he will never sound like you're in the same room.

Mega is announcing into a automatic (gamer??) headset or microphone and should probably stop. The volume changes up and down but that doesn't bother me. The noise level is high FFFFFFFFFFFF, but we can probably deal with that, too. It's an automatic microphone and that's deadly.

At the end where Mega stops talking and look what happens to the noise.

"Matt, how are you?" ........ffffffffffffffFFFFFFFFFFFF.

Pumping noise is impossible to correct. FFFFF is going to happen every time Mega stops talking to listen to Matt.

I have a headset where I can turn off AGC or Auto Gain Control. Mega should look for his controls.

I guess Matt is using a simple desktop microphone (or possibly built-in) and there is always going to be a presence difference between Mega and Matt. I once played two different people in a show by intentionally using a microphone that way.

There's no filter for that, either.

So Matt needs to change the room and Mega needs to change the microphone.

No problem.

OK. So I applied LF-Rolloff (or Low Rolloff) to both tracks.

Chris's Compressor to more or less even out the volume changes and then Limiter to take the peaks down a bit This is a badly cut "podcast" where I cut between the two tracks. It's in higher quality MP3 because 30 seconds of WAV will not fit.

If that's close enough, I'll forward details of how I got there.

Koz