Page 1 of 1

Normalizing?

PostPosted: Wed Sep 10, 2003 3:13 pm
by madmax
OK ... I was resampling a bassline last night (actually to put on the board - hopefully I'll get to it tonight) and realized that when I bounced it to computer via SCSI it wasn't normalized. So I looked at the sample that I had just bounced, cos I autonormalize when resampling through the inputs, and the EMU said that it was normalized (sample edit, tools 3, gain, +0 = normalize). So I normalized it on the computer, which def made the sample louder and sent it back to the EMU. It didn't clip - it sounded louder.

So my question is, why did I have to do this? Has anyone else noticed this? It would be a real pain to have to bounce each sample to the computer to normalize and then bounce back to get the loudest signal possible.

PostPosted: Thu Sep 11, 2003 3:28 am
by lost_sound
I have noticed that too. when I was editing all of my drumbreaks in recycle and doing scsi transfers to the emu I noticed that now I have really loud drum breaks and I am not able to sample any thing directly to the emu and get even remotly close to the loudness of the drums. Maybe the normalize in the comuter is much more efficent.

lost_sound

PostPosted: Thu Sep 11, 2003 4:59 am
by sampleandhold
i have actualy notice something like this in wave lab. i will normalize there and it actualy doesn't seem to normalize all the way to the top. i have actualy had to do 6 db boost just to get the highest amplitude to actualy touch zero.

also, you guys know about "the first six" right? basicaly what i do when i normalize something, i will actualy do another boost by six more decibles. it is said that the transiants resides in the first six db and those can actualy clip with out any appearant distortion.

perhaps when you normalize on the emu, you can boost it another six db before hearing distortion, and perhaps then you can get away with not having to normalize over and over again as you transfer from emu to computer.

i don't know, just an idea. try it out though and tell us if it works...

PostPosted: Thu Sep 11, 2003 3:02 pm
by madmax
Werd thanks for the feedback y'all. There was a point in time that I limited every sample and it looks like I'll be going back to that (FYI its with a soft knee and very small amounts of limiting - just enough to get the sample as loud as possible).

I've also realized that when using the filters (the shelving filter in particular) that there is a limited amount of headroom with the gains until you start to clip. Until this point though you can really blow out the sounds by bumping the gains up (not nessecarily just for creating different timbres, but for simply beefing that biatch up). I will have to run tests to see whether this has anything to do with the initial headroom on the sample (I'm assuming it does, which means that if using these, you might not have to normalize at all, as you're just gaining the fuck out of it in the EMU anyway).

PostPosted: Thu Sep 11, 2003 4:05 pm
by drayon
why normalize? it would be better to sample your sounds about -3dB so u have headroom spare incase u need to eq. If u normalize to full code and then eq, your gonna hit the wall pretty hard and things could sound ugly not to mention u would be able to shape the sound in the desired manner.

BTW EMU vs Puter.... both units use different convertors, your puter prolly has 24 bit convertors while the EMU has 18bit. The theoretical dynamic range of a 24bit convertor is much hight than an 18 bit convertor.
Internal mix bus resolution also plays a role in what u are experiencing. Both machines just calculate the range differently...also the EMU appears to have a little (4dB) headroom past what it reports as 0dB it then goes into like a soft clip zone before full on brick wall clipping.

PostPosted: Thu Sep 11, 2003 4:48 pm
by madmax
drayon wrote:why normalize? it would be better to sample your sounds about -3dB so u have headroom spare incase u need to eq. If u normalize to full code and then eq, your gonna hit the wall pretty hard and things could sound ugly not to mention u would be able to shape the sound in the desired manner.


Right, see what I wrote above about the gain on the filters. You're confirming that if using it, the sample headroom does make a difference and that it's not necessary to normalize when using these filters. However, when I'm using a resampled sound with only subtractive eqing, filtering going on, I'd prefer to have the hottest possible signal going into the analouge board so that I can have the lowest possible noise floor.

drayon wrote:BTW EMU vs Puter.... both units use different convertors, your puter prolly has 24 bit convertors while the EMU has 18bit. The theoretical dynamic range of a 24bit convertor is much hight than an 18 bit convertor. Internal mix bus resolution also plays a role in what u are experiencing. Both machines just calculate the range differently...also the EMU appears to have a little (4dB) headroom past what it reports as 0dB it then goes into like a soft clip zone before full on brick wall clipping.


No doubt, that's interesting about the soft clip zone.

The thing about recording to the computer, for many sounds, I actually prefer the colouration that the EMUs converters give. That being said, I do use the old 2408 for resampling as well. Especially when valuing clarity over colouration. There's another thread somewhere about this.

PostPosted: Fri Sep 12, 2003 1:21 am
by drayon
madmax wrote:
drayon wrote:why normalize? it would be better to sample your sounds about -3dB so u have headroom spare incase u need to eq. If u normalize to full code and then eq, your gonna hit the wall pretty hard and things could sound ugly not to mention u would be able to shape the sound in the desired manner.


Right, see what I wrote above about the gain on the filters. You're confirming that if using it, the sample headroom does make a difference and that it's not necessary to normalize when using these filters. However, when I'm using a resampled sound with only subtractive eqing, filtering going on, I'd prefer to have the hottest possible signal going into the analouge board so that I can have the lowest possible noise floor.


Ok, when set the EMU sampling mode to Resample....set it to 16 bit this lowers the head room an u should hit the ceiling. Or are u saying when u resample u are sending the signal out the EMU to an External unit, recording it then setting the EMU to sample from the source that contains the original sample that came from the EMU? thats to say are u not using the EMU's internal 'Resample' function?

drayon wrote:BTW EMU vs Puter.... both units use different convertors, your puter prolly has 24 bit convertors while the EMU has 18bit. The theoretical dynamic range of a 24bit convertor is much hight than an 18 bit convertor. Internal mix bus resolution also plays a role in what u are experiencing. Both machines just calculate the range differently...also the EMU appears to have a little (4dB) headroom past what it reports as 0dB it then goes into like a soft clip zone before full on brick wall clipping.


No doubt, that's interesting about the soft clip zone.

The thing about recording to the computer, for many sounds, I actually prefer the colouration that the EMUs converters give. That being said, I do use the old 2408 for resampling as well. Especially when valuing clarity over colouration. There's another thread somewhere about this.


This indeed is the beauty of the machine im so fond of as well. The unit has 'Attitude'. I agree with your overview on this as well. Coloration is great for many sounds like basses, drums etc but often for pads its nice to have let them flow freely out of the synth outputs so they remain uncolored or if they need to be sampled, record via some transparent device an cleared convertors an SCSI to the EMU. The EMU will still color the sounds but not the the extent of sampling the sound via the A/D's.

PostPosted: Fri Sep 12, 2003 2:51 pm
by madmax
drayon wrote:Ok, when set the EMU sampling mode to Resample....set it to 16 bit this lowers the head room an u should hit the ceiling. Or are u saying when u resample u are sending the signal out the EMU to an External unit, recording it then setting the EMU to sample from the source that contains the original sample that came from the EMU? thats to say are u not using the EMU's internal 'Resample' function?


It is rare that I use the internal resample. I could be wrong, but I'm under the impression that it coverts D>A and then A>D internally. If it applied destructive digital effects (filters, eq, amp env, pitch, etc.), I'd be more inclined to use it. However, as there are A/D conversions involved, I figure I might as well run it through the tube one more time (this is for most bass and drum sounds, which are usually the ones that you want to resample the most anyway). This way, the harmonic overtones that are lost in the conversions are (sort of) made up for.

drayon wrote:This indeed is the beauty of the machine im so fond of as well. The unit has 'Attitude'. I agree with your overview on this as well. Coloration is great for many sounds like basses, drums etc but often for pads its nice to have let them flow freely out of the synth outputs so they remain uncolored or if they need to be sampled, record via some transparent device an cleared convertors an SCSI to the EMU. The EMU will still color the sounds but not the the extent of sampling the sound via the A/D's.


I feel you 100%. Sometime the clarity of of the 2408 (24bit 96kHz converters) is nice!