Audacity and "compression"

[Retinend]

I want the sound wave to look like it does on the left (I did nothing to make it so).
At the moment one (human) speaker in the sound wave is much louder than the other.
Using "compressor" or "normalise" under "Effect" does nothing to bring about the desired equality of volume between the two (human) speakers.
I thought I knew what "compression" was: that it is what I was trying to do. But since "compressor" doesn't work, maybe I didn't and don't?
What do?

[the]

It could help if you explain what you've got there. Is it a song or speech?

Presumably it's just the perceived volume that changes at that point? Or does the character of the sound change in other ways as well?

And do you want to match the second half up just to make it more comfortable to listen to, or are you trying to 'repair' it and make the change undetectable (ie. for an audience, or use in an audio project)?



The initial thing to say is that you never want to make the waveform look the same, you want to make it sound the same. Do it by ear, not by eye (although there are of course visual readouts that will help).

[NoSleep]

Better not to use compression to try and level two voices that are completely different volume levels. Better to edit the parts, keeping them in position, then put one of the speakers on a new track and change the relative volumes between both tracks.

[Retinend]

It could help if you explain what you've got there. Is it a song or speech?

Presumably it's just the perceived volume that changes at that point? Or does the character of the sound change in other ways as well?

And do you want to match the second half up just to make it more comfortable to listen to, or are you trying to 'repair' it and make the change undetectable (ie. for an audience, or use in an audio project)?

Thanks for your help.

It's speech. An amateur radio show where the callers are way louder than the host.
On the left is an example where there is no difference, just for comparison.
Yeah I want it to be more comfortable to listen to.

@NoSleep: so I need to use audacity's "amplify" manually, to boost the host's volume to match the caller?

[NoSleep]

It would be better than using compression, which is designed to smooth out the minor changes of dynamics within the normal range of a single human voice. Trying to use a compressor to "flatten" the louder voice in comparison to the quieter is going to have a "pumping" effect on the louder voice.

If you simply make cuts wherever the voices change over and move all the quieter ones onto a new parallel track you will be able to adjust between the two until they sound more natural.

[Retinend]

You know that's a real time saver because for some reason, although the host talks a lot more, my intuition was to boost (amplify) his sections across the board instead of reducing the volume of the caller, who seldom talks.

So basically "compression" is something that makes an isolated vocal track sound nicer? Not much else?

And what is "normalise"?

[NoSleep]

Normalise looks at a sound file for the loudest peak and then boosts the file by the difference between the peak and 0dB (max).

[NoSleep]

So basically "compression" is something that makes an isolated vocal track sound nicer? Not much else?

It was originally used to reduce the signal to noise ratio when recording to tape; you can record to tape much louder if the dynamic peaks and troughs are brought closer together. Compression can also have a "sound" which people like (not just on vocals). In "modern" (the last 60-70 years) production, compressing vocals makes them sit nicer over the backing track. Compressing drums and other instruments can bring certain qualities out of a performance. It is easy to overdo it and good compression use generally sounds transparent. You can compress a whole mix or a group of instruments. In a way the interaction of several instruments or voices in a compressor is more interesting than compressing everything individually.

[GoblinAhFuckScary]

I know this might not be helpful, but have you considered just stealing a torrent of ableton or something? That Audacity compressor looks anything but intuitive

[Retinend]

Will Ableton do what I want it to do?

[GoblinAhFuckScary]

Will Ableton do what I want it to do?

Yes it will! Compression is just a way of reducing dynamic range to tighten a signal, giving it the flattened waveform you're after. Have a general look into eq, compression, limiting etc. Normalising is only really for mastering when you want to boost a signal to its optimum range before peaking.

Edit: Sorry if I'm just repeating what NoSleep explained

[Retinend]

Ok great! What is the function I need once I install it?

[GoblinAhFuckScary]

Ok great! What is the function I need once I install it?

Compressor. Use equaliser before compression and drop out frequencies that are significantly higher than other ones. Ableton should have a spectrum analyser in its eq so you can easily see the ugly noises

[NoSleep]

DON"T use a compressor to flatten both speakers to the same level. The louder voice will end up sound like it's "pumping". Edit them, like I said. Compression shouldn't be used lazily for fixing this issue. You can compress it to better effect after you have edited and adjusted the respective volumes of the speakers.

[GoblinAhFuckScary]

DON"T use a compressor to flatten both speakers to the same level. The louder voice will end up sound like it's "pumping". Edit them, like I said. Compression shouldn't be used lazily for fixing this issue. You can compress it to better effect after you have edited and adjusted the respective volumes of the speakers.

Ah, yeah do this first.