[sdiy] Digital accumulator VCO core?

Brian Willoughby brianw at audiobanshee.com
Tue Feb 16 02:15:30 CET 2021


Sub-sample alignment of the MinBLEP will create some variation. Also, if the engine runs at high resolution and then dithers down to 24-bit to avoid correlated quantization, that will add some randomness by definition.

I also assume that each step is potentially a different amplitude, and thus the MinBLEP would be scaled differently each time it's mixed it.

Brian


On Feb 15, 2021, at 17:09, cheater cheater wrote:
> I'd much rather prefer some randomness, so that the transitions don't
> all sound the same.
> 
> On Mon, Feb 15, 2021 at 11:35 PM Tom Wiltshire wrote:
>> 
>> The point as not to remove the aliases by *filtering* but by *cancellation*.
>> 
>> Imagine you could create a waveform that consisted of only “anti-aliases” that would eliminate the aliased harmonics when applied to the original signal. That’s what we’re doing.
>> 
>> Yes, the usual scenario is that the sample rate is fixed and therefore the band-limited data only needs to be derived once. You can do all the hard sums offline, generate one massive look-up table of the band-limited edge correction, and then you just paste the right bits into the right places and all the aliasing disappears. That’s why I described it as “magical”!





More information about the Synth-diy mailing list