Korg Forums

CharlesFerraro wrote:I believe the the RADIAS is 24 bit, but I wish there was a way to test this. If I gleaned any info off Axxim's post it seems the bit depth is a part in a greater whole that only offers a rough estimate(...?)

I'd imagine you could just see where the noise floor is, and the lowest possible audible signal from the Radias? (Although this would be difficult to discern against the noisefloor.)

16-bits go from 0dB (maximum audio signal) through to -96dB (the noise floor, although in practice it's higher [when recorded]), while 24-bits go from 0dB (max) to -144dB (noise floor).

You'll never achieve anywhere near a full 144dB of dynamics, but anything below -90dB or so might imply a 24bit output.

the source from the previous post:
http://en.wikipedia.org/wiki/Korg_RADIAS

Timo if you spare the time can you elaborate a bit on the relationship between bits and db?

I thought that bits and sample rate were determining the resolution of the "produced" waves prior to D/A conversion only

cibine wrote:the source from the previous post:
http://en.wikipedia.org/wiki/Korg_RADIAS

yeah i was saying that wikipedia article doesn't cite a source.

Most wiki articles have a bibliography type thing at the bottom.

this is just a guess but i think if you look at the other korg products of the same generation - m3, kronos, kaoss pad, they run at 24-bit 48khz.. so it might be safe to say that radias does as well. I am a bit of a noob when it comes about bits and hexa's but to my ears the radias sounds a bit more pristine than 16-bits

cibine wrote:this is just a guess but i think if you look at the other korg products of the same generation - m3, kronos, kaoss pad, they run at 24-bit 48khz.. so it might be safe to say that radias does as well. I am a bit of a noob when it comes about bits and hexa's but to my ears the radias sounds a bit more pristine than 16-bits

Good deduction. X-Trade also mentioned that. Going to try out Timo's suggestion too (that genius). Don't know how the noise floor works with the preamp though, it could be "dirty" and boost a lot of noise that's irrespective of the internal processing. Do I want the master volume on min or max? I'll try both and see what happens.

Like I keep parroting, we're talking about something that's irrelevant to the programming aspect of synths. You don't need to know stuff about "bits and hexa's" in order to patch a good lead, pad, or 8bit sound. Guys like Axxim are just super smart and go that extra mile. Doesn't hurt to know the nitty gritty but I totally derailed this thread by focusing on it.

foodeater wrote:

cibine wrote: What I like with the sound that comes out of the radias is that it almost makes a new 8 bit sound - you get the effect that resembles the 8-bit music but crispier, cleaner and with a really big harmonic range. Obviously the result is different though.

I'm with you there! I think radias is exceptional for "strange" digital sounds. I mean even younger kids who didn't grow up with 16-bit and earlier video games often find the sounds weird.

ps I like your track ear cleaner too. Is that a radias?

The ear cleaner was made with a monotribe.. i was just posting some tracks so i thought to add that even though it was not relevant to the 8-bit style...

CharlesFerraro wrote:...For some reason I thought if the waveform was going slow enough I could see the individual steps in the slew but I don't think the system works like that....

I supose you speak of looking that wave thru a oscilloscope or some similar? In that case you would need a scope that at least has a vertical screen resolution of 16.777.216 (2^24) pixels (and a super magnifying glass) until you can distinguish each bit change level

sorry if i go a little out of topic again, but I was missing such nice discussions in this forum for a while..

The difficulty to get 8 bit sounds with actual synths is that not only the waves are/were 8 bit but also all their modulators. The Resolution of their envelopes or LFOs had even less bits of resolution (in many cases only 4) so the steps of them where clearly audible.

Low resolution waveforms have their own characteristic due their steps. Each step is is a fast change from one state or value to other. The bigger this step is, the bigger are the overtones/harmonics it creates so there is a notable sound difference between a fine resoluted saw and a very steppy one. The first will have mainly the typical harmonics of a saw while the second will have a lot of additional ones.

Even the "newer" vintage synths (analog with MIDI) have their audible 8 bits due MIDI. The steps are clearly audible when the filter cutoff is modulated slowly via MIDI CC's and its resonance is set to a high value.

By the other side, 8 Bits are still present in 16, 24 or more bit systems: MIDI control values are still 8 bit and it depends on how they are internally processed. In such cases, even the finest D/A's wont help if there aren't good interpolation and filtering procedures to smooth those ancient 8 bit MIDI values that are the source for modulations.

Its funny to see how the evolution in sound/music reproduction has advanced from records and cassettes to CD's from analog to D/A systems with 16, 24 or even more bits, to end in the younger generations reproducing this improvements thru squeaky cell-phone speakers.

axxim wrote: Even the "newer" vintage synths (analog with MIDI) have their audible 8 bits due MIDI. The steps are clearly audible when the filter cutoff is modulated slowly via MIDI CC's and its resonance is set to a high value.

God I hate this step "glitch" so much! I wish that they can do something about the resolution- maybe its time for a "24-bit" MIDI or something.

The only thing I am aware of is that instruments with OSC implementation can provide higher resolutions in individual parameters. I don't understand though, a lot of VSTs seem to have decimal subdivisions on filters and other values - it would make sense for new hardware with more powerful DSPs to be able to give you more resolution even if its just from the onboard controls.

Thing is that really 7 bit (as used with MIDI - providing 0-127) is enough to accurately save a preset. I mean, you can't move a knob with 128 steps of precision anyway, most of the time, in an analog the driff and other inaccuracies of parameters is going to be close to that resolution.

8 bit resolution (0-255) would be plenty. The problem is moving between these values convincingly.
I know the Kronos, OASYS, and some other newer kit uses smoothing for the input values. Not sure if the Radias does this.

But we are talking about parameter resolution and not audio resolution there.