Should I normalize single-shot samples?

[superddman]

I am sampling one of my synths as a series of single-shot samples. Everything is fine except that the samples vary in volume.

Is it a good idea to normalize them afterwards (using "Maximize Peak Value" setting)?
If normalizing is a bad idea, what would you suggest to bring up their volumes?

[jupiter8]

Normalising is fine.The only problems is if they vary a lot in volume you might get differences in quality but you have to mess up quite bad for that to be noticable.

[superddman]

Normalising is fine.The only problems is if they vary a lot in volume you might get differences in quality but you have to mess up quite bad for that to be noticable.

Cool, cool

What about normalizing using "Maximize Peak Value" setting? Is that the correct way to do this?

[jupiter8]

Trying to think of a single reason not to and can't think of any. Maximise peak value should be fine.

[freeztar]

When you normalize the samples, it will raise the noisefloor as well. Hence, samples that have not been normalized will have less noise. You should be ok in this case, but it makes me wonder, how did you sample? Could velocity be the culprit?

[superddman]

Ok, thanks for your help. So I will definitelly normalize all samples afterwards using "Maximize Peak Value" setting.

On the other hand, should the velocity be set at the default level or should I maximize it when sampling?

Also, do you guys recommend some FX applied on the Master output when sampling?

[Charlie]

You could also record all the samples into one single audio file and then normalise the resulting big audio file before splitting into samples.

[jupiter8]

Ok, thanks for your help. So I will definitelly normalize all samples afterwards using "Maximize Peak Value" setting.

On the other hand, should the velocity be set at the default level or should I maximize it when sampling?

Also, do you guys recommend some FX applied on the Master output when sampling?

If you want to sample the sound at maximum velocity obviously you must have max. velocity. The sound is most likely different at different velocities.I'm not sure i know what you mean.

FX depends. Some,like chorus can sound like crap 'cause it will speed up/down with the pitchshifting. Does'nt have to be noticable but can be very disturbing. Delay would be pretty annoying since you'd have different delay times on different pitches.

You could also record all the samples into one single audio file and then normalise the resulting big audio file before splitting into samples.

What would the point in that be ? Then only one sample would be truly normalised and the rest not so.

[Sam@Megablastic]

You have to remember though that no instrument will ever be at max volume in a mix. I tend to mix each channel in Cubase 4 at -18db to -26db, and I still end up at around -2db on the mnaster fader.

If you record at 24 bit then there is no need to take it to the absolute maximum. A lot of a/d converters work better below -4 to -6 db any way.

[freeztar]

A lot of a/d converters work better below -4 to -6 db any way.

I haven't heard that before. Do you have more info or maybe a link?

[Sam@Megablastic]

I don't have the link, but it was a discusion on "gearsluts". It went something like that mid range cards sound better in that level range, and the top end cards like lynx etc can pull off better quality nearer the 0db mark.

But there really is no need to record so high, especially with with intersample overs. Some intersample overs can be as much as 8db, so it's best to play safe. With 24bit (probably 20bit in real world figures) it's safer to record with a good margin below 0db. It's not like the old 12 bit must sample as near to 0db to get all the resolution anymore 24bit has so much head room it's silly

Also try bring your levels in cubase on each track to -18 to -24 db, not the level fader, keep the level fader on the channel at 0db and lower the actual .wav or vsti at source. Your mixes will thank you for the inter sample overs not giving any plugins you use square waves. It also helps get rid of that "digital sound".

[superddman]

So I guess I won't apply any fx on the master channel when sampling. Also, I was thinking that the default velocity should give me the most natural sound when sampling. Maximum velocity can negatively alter the sound.

After researching a bit, I have came to conclusion that I will sample at 16 bit/44100hz since there is basicaly no audible difference with 24bit and file size is much smaller.

I am only talking about single-shot samples here. I think that the "head room" that Sam is referring to is mainly applicable to the entire mix, but not so much when sampling single-shot samples.

[Sascha Franck]

Are you planing to sample various velocity zones? In that case normalizing might be a bad idea.

[superddman]

No, just sampling one signle-shot samples with a default velocity.

I have just run some tests sampling at a default velocity and then normalizing peak value but I get lots of noise

Any way to avoid this?

How can the samples sound loud without normalizing them?

[JonathanDragonDJ]

which normalizing program will you use? i know cubase can do a non-destructive nomralizing...then you can just listen to it...and UNDO if you want to....if you can:t then just make a copy of the samples and normalize that....see how it sounds to you....best way to check! good luck