Audio Cleaning with Audacity and The Levelator

I was working on some very old video from a VHS tape when I discovered that the video and audio parts should be cleaned up separately. I was already using VirtualDub for the video part, so what about the audio? There are two great applications which both need to be used: Audacity and The Levelator.

My favorite audio editor is Audacity, and it’s excellent for recording and editing audio. Best of all, it’s free and open source, and it has an excellent Windows version. It’s exactly what I need. My only gripe is that it has its own audio format, so anytime you import audio, it takes the time to “import” it; and when you export, you actually need to choose your export format, whether WAV, MP3, or OGG, and do export processing– which can take some time.

Aside from that, it’s great for everything from chopping up audio to making some parts louder and some parts softer. It even has noise cancellation– but be extremely careful about over-applying the noise removal filter, because you can easily make the audio sound worse than it did to start with. Using Audacity to improve an audio file is very manual, and you’ll never get the whole thing to sound, as a whole, much better than the original recording.

Enter The Levelator. This is a new, free tool by GigaVox Media, and it comes in Windows, OS X, and Linux versions. It’s not open source, but it is free– for both non-commercial and commercial uses. The best part is that it’s totally automatic. No knobs, no settings, no checkboxes, no numbers, no nothing! All you do is drop your audio file into the app, and you get a .output file which just sounds better. According to the creators, The Levelator “adjusts the audio levels within your podcast or other audio file for variations from one speaker to the next, for example. It’s not a compressor, normalizer or limiter although it contains all three. It’s much more than those tools, and it’s much simpler to use.”

They have an excellent interview with the mathematical genius behind The Levelator, published as– of course– a podcast! If you have a spare 44 minutes, I recommend listening to the interview with Bruce Sharpe, the wonderful developer behind the software.

But if you don’t have time, here are some things I learned from the podcast. Also, I added a few of my own tips, since I spent a few hours playing with The Levelator:

1. Apply a little noise reduction first, but be very conservative. Never apply it 100%; even 50% is a good maximum. Really beware over-reducing the noise.
2. Raise the volume of extremely quiet parts, so The Levelator knows you actually want to hear those parts. Also, apply peak normalization.
3. Perform any editing you may want to do, such as trimming off the ends, cutting out certain parts, and moving stuff around (optional).
4. Throw it into The Levelator!
5. The Levelator can come very early in your process. You can do final, detailed editing afterward, and be sure to listen to all of it to see how it came out.

Why doesn’t The Levelator take MP3 files?
The Levelator is all about quality audio, and with its processing algorithms, it would need to decode MP3 to PCM or AIFF. Then, to output an MP3, it would have to re-encode the audio. This hammers the audio quality, because MP3 is designed to be encoded once. MP3 is a final step, and should never be intermediate (if you can avoid it).

In fact, if you decode and re-encode an MP3 file without making any other changes to it, it will sound worse. This is not the case with The Levelator; if you put the same file through over and over, it should come out sounding about the same. If you put in a perfect audio file, it will come out still perfect– that’s the design goal, anyway. And it makes a noticeable difference, especially on audio that’s poor to begin with. It makes it sound awesome!

So if you have a podcast, or even if you’re editing old VHS video recorded into digital (like I was), give The Levelator a try. Yes, I was quite impressed :)