[Resolved] Disaster Recording, Need Emergency Advice On Fixing Audio In Post

Posts about media, music and video
Post Reply
MavenOfMisfortune
Posts: 218
Joined: Fri Jul 24, 2015 10:13 am
Location: District of Columbia
Contact:

Fri Oct 09, 2015 3:11 am

While my first videos came out fine, some time between the second and third recording sessions my settings got thrown out of whack. Now I have almost 8 hours of content recorded I can't rerecord...that all have my voice too loud and the game sound too quiet. I need to know if there is any way at all I can fix this in post, and I was hoping someone here might know.

All the files were recorded as .mp4s through OBS so there's no multiple channels, and I don't have much money to spend on software. But I might need to spend money or hire a professional if that's the only way. This is some of my favorite work and it'll never come out the same. Any way of raising the game volume in the back or softening my voice would be a godsend and prevent me from uploading videos with super-imbalanced audio.

Please help. Don't know where else to turn. And I'm pretty desperate. I'm willing to take any and all options that might help adjust the sound, even if it makes my voice sound funny or warps it or anything like that. I need to know if there's something or if it is literally impossible. =(
Image
Image
ticktock1231
Posts: 419
Joined: Fri Feb 06, 2015 10:51 am
Location: Minnesota, USA

Fri Oct 09, 2015 8:32 am

Best of Vibes, I am not sure if you can adjust audio of a single channel. I have used Adobe Premier Pro to normalize volume. There are a few different movie edit software that is free that should give you the same basic audio normalization abilities.

Did you find out what the issue was? Were you able to fix it? I wish you the best of luck and write back with what you find out. I hope I am wrong about my thoughts on the audio and you can clean it up.
TeeAlarm
Posts: 3
Joined: Sat Oct 10, 2015 7:08 am
Contact:

Sat Oct 10, 2015 7:18 am

A Limiter/Compressor with the right settings should do the trick. Your voice will lose resolution in the process but you expected as much. You should be able to find it under effects, filters or plug-ins in your editor of choice.

Set it to attack everything above X db (make sure all game audio is below X). If you are using a plain (peek) limiter the second step is to normalize the track. Or you can play with the settings of a more complex compressor to achieve the same effect, i.e. the voice gets squished and everything including the game audio gets louder.
User avatar
VerveLight
Posts: 3
Joined: Thu Aug 20, 2015 5:07 pm
Contact:

Sat Oct 10, 2015 10:06 am

In Adobe Audition you can modify sound on a graph, but not sure if it'd help for you, you can for EG delete noise with it (such as a fan or rumble) but not sure if it can plain lower it.

http://tv.adobe.com/watch/companion-vid ... ing-noise/

About 5 minutes in.
MavenOfMisfortune
Posts: 218
Joined: Fri Jul 24, 2015 10:13 am
Location: District of Columbia
Contact:

Sat Oct 10, 2015 11:37 am

[QUOTE="TeeAlarm, post: 29136, member: 59988"]A Limiter/Compressor with the right settings should do the trick. Your voice will lose resolution in the process but you expected as much. You should be able to find it under effects, filters or plug-ins in your editor of choice.

Set it to attack everything above X db (make sure all game audio is below X). If you are using a plain (peek) limiter the second step is to normalize the track. Or you can play with the settings of a more complex compressor to achieve the same effect, i.e. the voice gets squished and everything including the game audio gets louder.[/QUOTE]

This is the post that has given me the absolute most hope. SO THANK YOU SO GODDAMNED MUCH YOU BEAUTIFUL CREATURE YOU. I'd kiss you if I could! I tried fooling around with settings in VLC (since it was the only program with a compressor that would open my video file) and IMMEDIATELY noticed the difference! My issue now is figuring out how to set up to sound best.

Keep in mind I'm a total newbie when it comes to editing stuff like this, so apologies if I ask a stupid question or don't know things. For instance, I have no idea how to tell what db my audio is from one point of the video to another. I supposed that Audacity might have allowed that, but I was unable to figure out how to export the audio in a format Audacity would both allow and not play in screeching static.

Currently, the settings on the compressor that have made things sound better are as follows ( following suggestions from https://lifeprotips.com/tip/204/compute ... crease-the ):

RMS/Peak: 0 (I have no idea what this does.)
Attack: 24.9 ms (Not sure if setting it lower would make things worse or better, so leaving it here for now.)
Release: 99.9 ms (Ditto.)
Threshold: -30 dB (Yeah, I know, probably excessive.)
Ratio: 20:1 (Felt like I needed it that high.)
Knee Radius: 9.9 dB (I have no idea what this does either.)
Makeup Gain: 24 dB (Felt like I needed it that high too.)

Any tips on how to proceed or figure things out from here?
Image
Image
TeeAlarm
Posts: 3
Joined: Sat Oct 10, 2015 7:08 am
Contact:

Sat Oct 10, 2015 2:42 pm

Most pro(sumer) video editing software should have a compressor module built in.
But if you are working with say Windows Movie Maker you can always split audio and video. On Windows I like to use a free software called XMediaRecode for that. You can than go on and edit the audio file in an editor like the free Audacity.

You can think of a compressor as a Robot turning the volume knob. Attack is how fast he turns the volume down once the decibel threshold is reached. Release is how fast he turns it back up, both in milliseconds. Ratio and Radius determine the amplification curve. It's how fast/far the robot turns while attack and release are more of a delay. Gain gives the volume back to the whole mix, be careful, if it's too high it will clip. In a post processing scenario I like to run a second normalize pass instead. These settings are modeled after physical devices that for example radio stations (ab)use to make everything louder and thus flatter and less dynamic.

I wouldn't go lower with the threshold than the game audio or you'll start compressing that too. In Audacity a quick way to tell is to just mouse over the waveform and look at the status bar at the bottom. And/or make the track take up (just drag it) the whole hight of the screen. That way you should get a feel of the level ranges in your mix.
Other than that you can really just experiment with the settings until it sounds right to you.
MavenOfMisfortune
Posts: 218
Joined: Fri Jul 24, 2015 10:13 am
Location: District of Columbia
Contact:

Sat Oct 10, 2015 3:07 pm

[QUOTE="TeeAlarm, post: 29160, member: 59988"]Most pro(sumer) video editing software should have a compressor module built in.
But if you are working with say Windows Movie Maker you can always split audio and video. On Windows I like to use a free software called XMediaRecode for that. You can than go on and edit the audio file in an editor like the free Audacity.

You can think of a compressor as a Robot turning the volume knob. Attack is how fast he turns the volume down once the decibel threshold is reached. Release is how fast he turns it back up, both in milliseconds. Ratio and Radius determine the amplification curve. It's how fast/far the robot turns while attack and release are more of a delay. Gain gives the volume back to the whole mix, be careful, if it's too high it will clip. In a post processing scenario I like to run a second normalize pass instead. These settings are modeled after physical devices that for example radio stations (ab)use to make everything louder and thus flatter and less dynamic.

I wouldn't go lower with the threshold than the game audio or you'll start compressing that too. In Audacity a quick way to tell is to just mouse over the waveform and look at the status bar at the bottom. And/or make the track take up (just drag it) the whole hight of the screen. That way you should get a feel of the level ranges in your mix.
Other than that you can really just experiment with the settings until it sounds right to you.[/QUOTE]

That's all fair, but my problem is that I can't figure out how to split the audio and video. Windows Movie Maker refuses to play nice with any files I make, so I use VLC usually for just plain playing and testing videos and either ffmpeg or Avidemux for video editing. VLC had the compressor I found. If I could get the audio into Audacity, I think I could figure out the dB levels pretty easily.

I still don't understand what you mean by normalize passes. But if ratio and radius are fast/far, then I'd want a high ratio and a decently high radius. I'll probably need a decently high gain, too, to bring the game sound up to more audible levels. So I'd say the main obstacle then is figuring out a way to get my .mp4's audio into Audacity without it reading the file as a three second clip of screeching.
Image
Image
TeeAlarm
Posts: 3
Joined: Sat Oct 10, 2015 7:08 am
Contact:

Sun Oct 11, 2015 3:21 am

Normalize just amplifies the audio to -1 db. (industry standard for max volume, you can amplify to 0db but that will clip on some devices, I have seen it do that mostly on older home stereo stuff, but it doesn't hurt to keep this small safety margin) It is (almost) a non destructive edit. Since compression is for the most part volume reduction I like to do the amplification as a second or last step, that is all. Just to make sure I'm not wasting db. Or in this case also to mitigate the risk of over amplifying with the gain settings.

You can use ffmpeg on the command line to extract the audio in .wav format. Or you can use the ffmpeg front end I mentioned earlier, XMediaRecode. There you would set output to custom, audio only or deactivate the video in its tab and set audio to WAV for the uncompressed windows audio file format.
MavenOfMisfortune
Posts: 218
Joined: Fri Jul 24, 2015 10:13 am
Location: District of Columbia
Contact:

Sun Oct 11, 2015 7:55 am

Ohh! Well, I'm glad you explained it! I found an option for that through Audacity so it seemed pretty effective. I slightly reduced the amplification afterwards, though, just to be safe. And yeah, I ended up using a combination of ffmpeg and Audacity to work it out. ffmpeg doesn't have the greatest UI, but I've been learning it best I can.

Thank you so much for your help, TeeAlarm! Thanks to you pointing us in the right direction, I was able to salvage all of my recordings so far. There's a little bit of audio glitchiness, but I expected it wouldn't be perfect given the original file. All I've got left to do is my normal cutting into 10ish minute segments via ffmpeg and then learn how to join two video files through it for part 11 and the finale parts.

You definitely can expect a thank you in the videos. ^_^
Image
Image
Post Reply