Noise in 8-bit samples (.mod)

Started by Claypool, March 08, 2007, 19:44:52

Previous topic - Next topic

Atlantis

Quote from: "Jojo"
Quote from: "Snu"
Quote from: "Sam_Zen"
Quote from: "Jojo"be sure to Normalize them first (the icon next to the note icon in the sample editor). this will reduce the noise a bit because you have a greater level and this results into more details
Sorry, I don't agree on this. Normalization means amplification, as much as possible without clipping, so if there's any noise in the sample, the noise will be amplified too.

I checked this mod-file, and it is strongly the case here. These samples are of very poor quality (8-bit or not) and have lots of noise in them. And they have a length which makes no sense. As you can see on the screenshot (sample 4 : HH-Op.wav), the actual sound is already faded out before the arrow. What's left after the arrow is just noisy nothing (it should be a clean zero-line).
Some remedies :
~ First remove the noise in a wav-editor with some noise-reduction tool, before using them in a tracker module.
~ Crop the sample just to the point where the sound has faded out completely (the 'arrow-point')
But : Search for some other drum-samples. These ones are really not good enough for professional purposes.

im thinking he suggested the normalization because normalizing it before its converted to 8bit will increase the amplitude before the majority of the noise floor is introduced (as opposed to the tracker changing the volume after the noise floor is introduced).
good idea on the cropping, i shoulda thought to mention that!

yes, that's what i mean... if you have a 16bit sample which is NOT noisy you should go and amplify it BEFORE you convert it to 8bit...
Yes, and before you do that you should make sure you're maintaining the high bit depth process and applying 8 bit dither at the end. As Sam_Zen said, it's better to do this in an audio editor, using a plugin like iZotope Ozone to apply dither rather than truncating the sound in OpenMPT.

Sam_Zen, just remember that attentuation adds quantisation noise too if the process is handled in 16 bit, and if you turn the listening volume up on the attenuated signal, you'll still hear the same noise floor as with the amplified signal. What I'm trying to suggest is to mask the quantisation noise with dither, but of course it's a good idea to get the original sample as clean as possible before that.
Put an end to the loudness war. Don't limit or compress your mixdown until mastering; leave the master channel alone.

http://www.facebook.com/atlanteanrecords

Sam_Zen

Quote from: "Atlantis"if you turn the listening volume up on the attenuated signal, you'll still hear the same noise floor
You're right, but that imo is part of 'post-production' circumstances. If one burns the tracks of an audio-cd on a too low volume level, so one has to amplify the tracks too much. Of course artifacts will occur.
This is about the preparations of the material before starting to compose, and, quite important, the right order of steps in that process.
0.618033988

Saga Musix

i think modplug did a better conversion from 16 to 8 bit when i tested it some years ago than audio programmes.. but maybe i've just chosen the wrong programmes :D
» No support, bug reports, feature requests via private messages - they will not be answered. Use the forums and the issue tracker so that everyone can benefit from your post.