To dither, or not to dither?
- KVRist
- Topic Starter
- 316 posts since 9 Jan, 2023
I remember watching an instructional Waveform video back about a year ago when I first started using the software, and the guy in the video suggested leaving it on. I never really gave it much thought, but now that I'm more comfortable with Waveform, I'd like to milk every drop of honey out of it that I can.
It's my understanding that most engineers suggest leaving it off, unless exporting down to a lower fidelity. My finished songs are really only used for Youtube, with the occasional upload to Soundcloud. Would adding that extra noise from dithering be more beneficial with respect to Youtube videos?
It's my understanding that most engineers suggest leaving it off, unless exporting down to a lower fidelity. My finished songs are really only used for Youtube, with the occasional upload to Soundcloud. Would adding that extra noise from dithering be more beneficial with respect to Youtube videos?
- KVRAF
- 15279 posts since 8 Mar, 2005 from Utrecht, Holland
It depends, but it's really simple:
- Export to 24 or 32bit WAV: dithering noise at -144dB is totally futile. You might leave it on, or off, no one will know.
- Export to 16bit WAV: better dither. But with a single-digit dB LUFS master: no one will know.
- Export to 8bit WAV: bro, why...
- Export to MP3: it's complicated (and I can explain but I'll charge for the lecture) but here dithering won't do much.
We are the KVR collective. Resistance is futile. You will be assimilated.
My MusicCalc is served over https!!
My MusicCalc is served over https!!
-
- KVRian
- 1097 posts since 28 May, 2010 from Finland
Electronically speaking you should always add dither, when you convert from one bit depth to another. The conversion process is numerically inaccurate and will produce low-noise distortion, which the dither helps to fix. However, some argue that its not always audible (while claim that the dither signal is). Perhaps this is why they allow one to not apply it. But technically speaking it should be put in by default.
- KVRAF
- 1804 posts since 23 Sep, 2004 from Kocmoc
Because it sounds better with fuller samples like pads. Do not dither bass samples when using 8bit.
I mainly do this for samples going to Amiga, for use with Protracker. Also for saving space on Fasttracker...
Really depends on the sound itself though in the end.
Soft Knees - Live 12, Diva, Omnisphere, Slate Digital VSX, TDR, Kush Audio, U-He, PA, Valhalla, Fuse, Pulsar, NI, OekSound etc. on Win11Pro R7950X & RME AiO Pro
https://www.youtube.com/@softknees/videos Music & Demoscene
https://www.youtube.com/@softknees/videos Music & Demoscene
-
- KVRian
- 1097 posts since 28 May, 2010 from Finland
If you want decimation effects then use a bitcrusher, not export settings. Just notifying people who might think of using export settings for "sound".legendCNCD wrote: ↑Wed Feb 14, 2024 1:24 pmBecause it sounds better with fuller samples like pads. Do not dither bass samples when using 8bit.
I mainly do this for samples going to Amiga, for use with Protracker. Also for saving space on Fasttracker...
Really depends on the sound itself though in the end.
Last edited by soundmodel on Wed Feb 14, 2024 1:32 pm, edited 1 time in total.
- KVRAF
- 1804 posts since 23 Sep, 2004 from Kocmoc
What? The target is 8bit/variable hz for this machine.soundmodel wrote: ↑Wed Feb 14, 2024 1:29 pmIf you want decimation effects then use a bitcrusher, not export settings.legendCNCD wrote: ↑Wed Feb 14, 2024 1:24 pmBecause it sounds better with fuller samples like pads. Do not dither bass samples when using 8bit.
I mainly do this for samples going to Amiga, for use with Protracker. Also for saving space on Fasttracker...
Really depends on the sound itself though in the end.
Soft Knees - Live 12, Diva, Omnisphere, Slate Digital VSX, TDR, Kush Audio, U-He, PA, Valhalla, Fuse, Pulsar, NI, OekSound etc. on Win11Pro R7950X & RME AiO Pro
https://www.youtube.com/@softknees/videos Music & Demoscene
https://www.youtube.com/@softknees/videos Music & Demoscene
- KVRAF
- 5512 posts since 2 Sep, 2019
Dithering doesn’t really “fix” quantization error, so much as it buries it in low-level noise.
And you’re not going to hear this quantization error or low level noise until the very end of a long fadeout. And that’s only if you have it turned up pretty loud. It’s mostly much ado about nothing either way.
And you’re not going to hear this quantization error or low level noise until the very end of a long fadeout. And that’s only if you have it turned up pretty loud. It’s mostly much ado about nothing either way.
Last edited by jamcat on Wed Feb 14, 2024 7:36 pm, edited 1 time in total.
THIS MUSIC HAS BEEN MIXED TO BE PLAYED LOUD SO TURN IT UP
-
- addled muppet weed
- 105882 posts since 26 Jan, 2003 from through the looking glass
- KVRist
- Topic Starter
- 316 posts since 9 Jan, 2023
Hmm...all right. Let me ask you this, then. So, all of my acoustic tracks are recorded on an old Roland VS-880EX. I record them at 44.1khz because otherwise, for whatever reason (age, most likely), I can't always successfully back up the VS data (not WAV files -- it's an exclusive format, I think, Roland used to use) to CD-R, which I need to do in order to bring it over to my PC.soundmodel wrote: ↑Wed Feb 14, 2024 11:11 am Electronically speaking you should always add dither, when you convert from one bit depth to another. The conversion process is numerically inaccurate and will produce low-noise distortion, which the dither helps to fix. However, some argue that its not always audible (while claim that the dither signal is). Perhaps this is why they allow one to not apply it. But technically speaking it should be put in by default.
I then use a homemade file that allows me to then convert that data in Reaper to WAV files, which are then imported into Tracktion Waveform for mixing.
Since Waveform uses 48khz, do you think dithering when converting upward has the same effect? Or would it just be when converting downward in fidelity that dithering becomes truly useful?
- KVRAF
- 15279 posts since 8 Mar, 2005 from Utrecht, Holland
You can stop right there: acoustical recordings. They always contain some noise (from background or preamps or whatever) which acts similar to dithering noise you'd add.
We are the KVR collective. Resistance is futile. You will be assimilated.
My MusicCalc is served over https!!
My MusicCalc is served over https!!
-
- KVRAF
- 6427 posts since 22 Jan, 2005 from Sweden
...forget about dithering for sample rate conversion. It's purely for bit depth from 24 to 16 bit and similar, 32 float to 16 and such.
In this case every 256 levels in 24-bit become 1 level as 16-bits.
- this truncation creates a pattern for brain to perceive as a frequency
- so you add noise to make this a bit random so not truncated at same level every sample
- less of a pattern for brain to perceive as frequency
How to add noise, and what type, depends a bit on music content. So there are different algos for this.
I use Waves L2 and while mixing it is set to 24-bit.
- for export I set to 16 bit so noise is added to the right level
- and type 1 algo, no shaping
Which type and if to use noise shaping is over my head currently. Maybe somebody knows how this works and for what musical content.
- KVRist
- Topic Starter
- 316 posts since 9 Jan, 2023
-
- KVRian
- 1097 posts since 28 May, 2010 from Finland
Pro-tip:
Create your own dither noise. E.g. by lifting the background noise of recordings or by using noise oscillators. On some silent acoustic recordings an artificial dither may sometimes not be audible as such, but when it is mixed with the room noise and it creates phaser effects.
Create your own dither noise. E.g. by lifting the background noise of recordings or by using noise oscillators. On some silent acoustic recordings an artificial dither may sometimes not be audible as such, but when it is mixed with the room noise and it creates phaser effects.
-
- KVRAF
- 2590 posts since 19 Mar, 2008 from germany
In general, almost always: Yes, it is unnecessary!
In particular, there is one exception: If you are making acoustic recordings,
e.g. an acoustic guitar playing alone, or an orchestra. And during the
recording there are places of calm, so that the guitar or the orchestra slowly
fades out and transitions into this calm, then dithering could make sense.
But only if you downsample from your 32-bit or 24-bit recording format to
the 16-bit CD format.
In all other applications, the overall dynamics of a recording are always so
small that dithering makes no sense at all.
free mp3s + info: andy-enroe.de songs + weird stuff: enroe.de