Fixing up the audio in a video

Normally when you record audio, the volume is not at 100%. This is to ensure that the audio doesn't clip. What this means is that after the video is recorded, you usually need to normalize the audio to make the audio louder. You could also add an audio compressor, but usually it's just enough to normalize the audio.

To make matters worse, there's echo, reverb, noise, etc.

This is not a trivial thing to fix - you might need to guess and check the audio for each video. And getting rid of reverb has been impossible until new AI tools came onto the scene.

You also don't want to re-encode the video files because that takes a long time and will degrade the video quality. Fortunately there's a few ways of doing it really quickly. Just pick one of the ways below.

Adobe Podcast

Adobe Podcast is a game-changing freemium tool that uses AI to basically recreate spoken audio. It works better than Audacity's OpenVino, Ultimate Vocal Remover UVR5 (haven't tried BS-Roformer), Supertone Clear (Goyo), etc.

Basically the AI in Adobe Podcast re-creates the voice completely - removing all of the background noises, echo, and reverb. This can result in some strange things where if there is too much noise in the background, the voice doesn't make sense because the AI guesses the syllables - where a human can actually work out the words.

After re-creating the voice, it then masters the audio by applying reasonable EQ and compression (although it does seem to be a bit bass-heavy).

If you have a video, just upload it here: https://podcast.adobe.com/en/edit-with-adobe-podcast

Or if you have extracted out the audio (e.g. use Audacity to separate audio from the video), simply edit it using https://podcast.adobe.com/en/enhance and then remux it using Xmedia Recode (see below).

You can use Adobe Podcast to delete individual words because it uses automatic transcription to identify individual words.

The free version doesn't allow you to mix back in the original audio so you may want to export the audio back out, then mix in some of the room sound, otherwise it can sound very unnatural - for instance you'll lose any laughter or clapping from the audience.

Modified MP3GainGUI using AACGain

https://www.videohelp.com/software/AACGain

If all you need to do is just normalise the gain of a video file, then AACGain is the easiest way. AACGainis a command line tool that can normalise the volume of AAC files. Since MP4 usually consists of an AAC audio track, you can usually normalise the audio using AACGain. 

By wrapping this up in the MP3GainGUI, you can have a really simple interface that just works.

I've done this here in the file below - just unzip and run MP3GainGUI.exe - it's a portable app! 

Definitely the easiest way to normalise the video of a file.

XMedia Recode

XMedia Recode can also batch convert files. You have a lot of control over each setting, and it is a little complicated. But it does work.

https://www.xmedia-recode.de/en/help/afilter/filter-volume-correction.php 

Here are the steps - you can bulk add files and have them all process one after the other. Basically, uou set the video to just Copy, while the Audio track you re-encode while applying the Volume Normalization filter.

FFMPEG Batch Converter

https://ffmpeg-batch.sourceforge.io/ 

FFMPEG can a simple method to normalise the audio, called loudnorm. By using FFmpeg Batch Converter as a GUI, you can just enter in the command line and run it across multiple files.

Just add the Parameter:

af loudnorm=I=-23:TP=-1.5:LRA=7

Shutter Encoder

Shutter Encoder has the ability to normalise audio.

https://www.shutterencoder.com/

It is a super advanced too for video encoding aimed at video professionals. But be warned it also might re-encode the file, which means the video file may lose quality and take a long time - you have to be very careful with the settings. I haven't tried this because it really does a lot.

But you can do other cool things like burn in the subtitles.