Scope/SC platform 5.0

[Shroomz~>]

I'll post a module, but I can't do it right now because sharc's using the sdk.

[hifiboom]

sounds interesting. i`m curious.

[Shroomz~>]

Since Hifi asked, here's a simple method/trick to downsample an audio signal in sdk from for example 44.1 to 22.5 kHz (or 11 or whatever). Nothing special & as I mentioned further up the page, it's not really all that relevant to what Hifi was talking about in terms of upsampling to high sampling rates & back down again. It's nothing impressive at all & would need the internal pot removed in use to reduce size, but anyway, here you go -

[hifiboom]

okay thanks shroomz.

But thats only half of the story, as the audio chain still runs at its native 44100 samples per second.

a "real" downsampling would calculate only 22050 samples per second and so save half of the scope dsp power or spoken in other words can calculate two 22050 samples-bitstreams at a time for the price of one 44100 sample-calculations.

otherwise every lowpass filter can be called a downsampler.

[Shroomz~>]

As I just said.

as I mentioned further up the page, it's not really all that relevant to what Hifi was talking about in terms of upsampling to high sampling rates & back down again.

[hifiboom]

[Shroomz~>]

I don't understand the point of the handshake imagery hifi. DSP to PC handshake?

[hifiboom]

no, the picture does describe "agreed".

As I just said

[kylie]

another form of 'fo shizzle ma nizzle ?

[Shroomz~>]

Ok Hifi.

It actually looks like a 'secret' handshake like something from a mason handshake teaching class.

[Warp69]

the lowest sampling rate is 44,1khz I guess, so you first upsample to a rate like 192 or 384, then do a process, and then downsample to 44,1 again.

So I think a more advanced split of the power and the possibility for internal sampling rate changes could be quite cool.

for example there is no need to run the basic stuff like mixers, etc. at much higher Sampling rates.

It would ofcourse be nice to have up/down sampling as modules, but would it work? You'll introduce x sample latency - depending on the algorithm ofcourse - when you downsample - better quality more latency. Let say you have a filter and a summing engine in some kind of series (and you're unable to combine those two functions), both using a up/down sampling module - the downsampling algorithm use ex. 11 samples of latency, so you would introduce 22 samples delay in your device compared to all other devices in the environment - we cant have ADC, since we have a realtime system. How should that work?

We could eliminate the latency, but that would introduce aliasing in the downsampling - the exact same thing we want to prevent in the first place.

I think nobody would downsample to 22khz and then upsample to 44,1 again as this would directly affect the audio quality in a bad way.

I would need to downsample to ex. 16kHz, if I choose to emulate the AD/DA stage of vintage reverbs.

[Shroomz~>]

Warp, why would introducing a latency like a 22 sample delay in a device be a problem if users are informed of it? Surely users can easily compensate for that if they're made aware of it? I've read that TD on Scope has a significant latency, but I don't have it so couldn't say exactly how much.

[Warp69]

Well - you could experience phase problems if you decide to mix the processed signal with the dry signal. But you're correct people could just compensate manually - they already have the problem with the look ahead function in a compressor as you descripted via the TD. But let say you have a path with sereval plugins with latency - combined around 200 samples - then that channel is delayed 4.5ms compared to all other channels from the DAW - you would have to compensate for every channel.

[Shroomz~>]

I guess that's what having a 0-200 sample delay per mixer channel is for.

[Warp69]

I guess that's what having a 0-200 sample delay per mixer channel is for.

Im also using that for HAAS effect

[Shroomz~>]

Good point..