Does Live 5 normalize when consolidating?

Tarekith · Post by **Tarekith** » Wed Jul 12, 2006 10:51 pm

I remember this was an issue with past versions of Live, does version 5 still do this?

Johnisfaster · Post by **Johnisfaster** » Wed Jul 12, 2006 11:08 pm

yes, for some odd reason it normalizes and then brings it back down to the original volume.. weird huh?

timothyallan · Post by **timothyallan** » Thu Jul 13, 2006 5:16 am

Rumor has it that Robert posted a reason as to why it has to do this somewhere, but i can't find the post...

Tarekith · Post by **Tarekith** » Thu Jul 13, 2006 10:56 am

If anyone can find this post, I'd love see the reason.

Tarekith · Post by **Tarekith** » Thu Jul 13, 2006 11:31 am

Ah, here it is:

Hi folks,

well, I think it's not a bug, it's a feature (sorry, love to say that Smile )

In Live a consolidated clip will be normalized (I explain later why). To compensate the volume change trough the normalizing the clip's gain is adjusted.

Example:
----------------------------
1) Put an audio clip into Live's arranger.
2) Select a small part in the middle of the clip and consolidate it
3) Now you have 3 clips. You will notice that the waveform view of the consolidated clip looks different as the same part in the original clip before.
4) But now compare the gain settings of these clips. The gain of the consolidated clip is lowered to compensate the normalization.

So because the consolidated clip is normalized, you see the a different amplitude in the waveform as before.

But why the normalize?
----------------------------
This was already an issue during the beta period of Live 4.0. So I will quote something Bernd already wrote concerning this issue:

Quote:

Here is a misunderstanding. Live does *not* change the relative amplitudes through consolidate. *But* we need to change the clip volume *because* Live makes quite an effort so that consolidate does change as less as possible in terms of the resulting audio.

So why do we need to change the volume? Some examples:

1) We have a clip with a sample that is normalized means it uses the whole resolution that lets say a 16 bit sample file offers. And now you raise the gain of that clip. What happens? That clip creates audio that is above 0 db. If we would write this to a file it would result in a sound file with clipping. So we need to normalize the audio before writing it to the file. But now we need to lower the gain of the consolidated clip to ensure that this clip will not create loader audio.

2) Internally the audio engine calculates in 32 bit float resolution. So every convertion of a intermediate signal to a 16 or even 24 bit resolution - and that is what happens when you consolidate - will result in a reduction of resolution - by definition, no matter of you use dithering or not. So if you now consolidate a very quiet part of a sample and Live would write it back to the file as it is, the sound would even loose more resolution.

So for all this and some other reasons Live always normalizes the audio before writing it to disk when consolidating to keep as much resolution as possible. This can result in a clip that has a gain different from the original clip, but again, it never results in a different amplitude of the resulting audio.

Kind regards, Bernd.

regards,
Alex

timothyallan · Post by **timothyallan** » Thu Jul 13, 2006 11:37 am

Nice job Tarekith

TheAnimal · Post by **TheAnimal** » Thu Jul 13, 2006 11:56 am

Thanks, Tarekith!

icedsushi · Post by **icedsushi** » Thu Jul 13, 2006 2:30 pm

I've always noticed that my clips after consolidating are adjusted to the proper volume to match wherever that track's fader is set. I never quite understood why this happens but it seems to match perfectly every time if I am mixing levels and creating consolidated clips at the same time.

If things are already at the right level, it's nice to have the clips adjusted relative to each other for you.

Maybe I'm not clear what I want to say, but anyway, the point is that I like this behaviour!

Tarekith · Post by **Tarekith** » Thu Jul 13, 2006 2:39 pm

It wouldn't be bad, but normalizing to 0dBFS should be avoid.