Audio Normalizing

[dvddemon]

I have a lot of music video's that are in different formats, avi, mpeg, mov etc. They all have different audio levels and I am trying to find a program that will normalize the level between files. I'm not looking for compression that will soften loud parts and increase soft parts. Just something that keeps the volume the same when going from file to file. Any suggestions? For example VLC has a normalize feature but it doesn't seem to work. I am looking for something similar to iTunes soundcheck.

[budwzr]

Normalization is destructive to the music. It flattens the dynamic range. The professional way to squash peaks is to use a compressor BEFORE mastering.

Having said that, I would recommend converting everything to high bitrate MP3's or AAC's and use a software like MP3Gain that has a "normal" volume reference point, and instead of normalizing, it uses the MP3 container's "Playback Volume" flag to adjust volume, and it does batch jobs.

[dvddemon]

Thanks,

But unfortunately I have almost 2000 video clips, so I was hoping to find a program that would normalize on the fly. Like as simple as checking a normalize checkbox in the software. Do you have any experience with the VLC normalize option? I agree that MP3 Gain is the way to go to get the highest quality, and I may slowly do that over time. But for now a quick solution is what I'm hoping to find. Thanks again

[budwzr]

Do you have any experience with the VLC normalize option?

No, I don't. I never use EQ's, bass boosters, etc. either.

[magillagorilla]

You would have to re-encode or remux all the files to change the audio tracks. Avisynth has a Normalize() function I use all the time, but my video is usually people talking.

Normalizing just finds the loudest db level in the clip then raises the total volume of the sound, equally, so that the loudest portion is at 0db. So if the loudest peak value in the clip is -12db, normalize would raise the total volume 12db across the entire clip. So, in that sense, you are not harming the dynamics, soft sounds verses lound sounds, of the audio. You are simply mazimizing the volume of the sound without clipping. Normalizing shoud not be distructive.

However, normalizing may not generate the continuity of "volume" you seek between your clips. Normalizing is limited to the loudest sound of the audio. So if you have a recording of a super quiet orchaestra -20db, but in 1 part of the song there is a cymbal crash peaks at -4db, normalize will only raise the volume 4db. Rock music is often squished for loudness already so if peak values are constantly at -5db you'll get a small volume pump.

A compressor limiter will actually change the db level of the sound throughout the clip dynamically. Quite sounds become louder and loud sounds are dampened. This technique will "squash" dynamic range and force all the sound in the clip to a target db level. This is generally a bad thing to do to music. Compression, as budwzr points out, is best reserved for the mastering stage of recordings.

[Cornucopia]

Playback Gainchange metadata (such as exists in AC3, DTS, and SOMEHOW in mp3 with MP3Gain) are non-destructive.
And if you apply normalization to a LPCM WAV file with EXACT amounts so as to create a simple BIT-SHIFT (and one where you had the headroom with which to do this), those can be non-destructive.
ANY OTHER KIND of Normalization is destructive/lossy, even for LPCM.

For constancy in "listenability", you would want to compare the LONG-TERM RMS average levels of each of your clips and match those (this is what AC3's Dialnorm intends to do, but many consumers don't understand it and work counter to it). This could be done with Playback Gainchange metadata, or with a smarter form of baked-in Gainchange.

...OR, you could just do what I do and manually adjust the volume control on your receiver...

Scott

[budwzr]

Well anyway, the OP wants to "volume average" between files, not in the waveform.

I have little experience with normalizing a waveform, so my nomenclature may be off. I suppose it WOULD be a useful thing for speech recording where the mic moves around and levels fluctuate.

[magillagorilla]

Normalization is really usefull on voice files. When I used to record music I would often normalize individual tracks before mixing. Technically there isn't effect or filter you can use on an audio file which is non-destructive. Change the waveform and you have altered the file, thus taking it to a non-original state.

There may be media players with auto-gain. That would be really easy and save a bunch of time.