signal types and ranges

unkargherth · 27 May 2005

lg wrote: What are the normal ranges of the various types of float and integer signals in SynthMaker?

• audio signals (are they -1 to +1 like in VST plugins? ) • frequencies (Hz? a fraction of the Nyquist frequency?) • pitches (MIDI note numbers?) • ADSR inputs and other times

• audio signals : as VST -1..1 • frequencies: unless noted 0..1 representing 0..Nyquist • pitches : More or less. Float value corresponding to MIDI Notes (0.0..127.0) • ADSR inputs: Atatck, Decay and Release in number of samples ( you'l find a nice expoential conversion from 0..1 inside VCA module in example.osm). Sustain 0..1. Please detail what "other times" you mean

.... and welcome here lq!!

stefan · 27 May 2005

lg wrote: What are the normal ranges of the various types of float and integer signals in SynthMaker?

• audio signals (are they -1 to +1 like in VST plugins? ) • frequencies (Hz? a fraction of the Nyquist frequency?) • pitches (MIDI note numbers?) • ADSR inputs and other times

Audio signals are in whatever range you want, but -1 to +1 is a good target range.

Frequencies are from 0 to 1, basicly it's normalized frequency, where 1 equals nyquist frequency. (If this seems incorrect, freq 0.5 is nyquist, but I'm quite sure 1.0 is 🙂)

Pitches are supposed to be equal to midi pitches, which means that pitch 60.0 equals middle C, or C-4, which equals approx 261.6Hz.

ADSR Inputs are in number of samples, so one second would equal 44100 samples, if 44100 is the current sample rate. The sustain input however is an ordinary float between 0.0 and 1.0.

I hope this answers your questions. Cheers!

edit: wtf unkargherth 🙂

bones · 27 May 2005

What is Nyquist Frequency? Didn't he write Signal To Noise and A Signal Shattered? [Good books, BTW]

skOre · 27 May 2005

Nyquist is half of the sample frequency.

Think of a waveform displayed. Think of the Line in the middle.

The whole thing covers 44.100 hz, so that is 22.050 from the middle to the top and another 22.050 from the middle to the bottom.

If something exceeds this, it will clip.

[dunno if this is correct 😉]

Wikipedias explaination

unkargherth · 28 May 2005

bones wrote: What is Nyquist Frequency? Didn't he write Signal To Noise and A Signal Shattered? [Good books, BTW]

This is Eric S. Nylund, not Nyquist 🙂

grease · 28 May 2005

skOre wrote: Nyquist is half of the sample frequency.

Think of a waveform displayed. Think of the Line in the middle.

The whole thing covers 44.100 hz, so that is 22.050 from the middle to the top and another 22.050 from the middle to the bottom.

If something exceeds this, it will clip.

[dunno if this is correct 😉]

Wikipedias explaination

Seems you're getting bit depth mixed up with sampling frequency - the sampling frequency just represents how many sample points are calculated each second. When the sampling frequency is at 44100 Hz (Hertz means "per second"), then according to Nyquist, the highest frequency that can be represented in the digital audio is at 22050 Hz (which is beyond the range of human hearing). This frequency is called the Nyquist frequency. I guess there's a lot more complexity to the story, with things like aliasing, etc. (I don't know that stuff too well!)

The height of the signal above and below zero is broken up into pieces according to the given bit depth. 16 bit audio has 2^(16) different amplitude levels. Any signal that has an amplitude above that will clip when digitized.

By the way, if frequency values are set with 0..1 representing 0..Nyquist frequency, then this seems to be dependent on what samplerate is being used. Or does the "Nyquist frequency" in SynthMaker always mean 22050 Hz?

unkargherth · 28 May 2005

grease wrote: By the way, if frequency values are set with 0..1 representing 0..Nyquist frequency, then this seems to be dependent on what samplerate is being used. Or does the "Nyquist frequency" in SynthMaker always mean 22050 Hz?

That's what y say 0.. Nyquist and not 0..22050. AFAIK (and tested , using asio at SR 48000 ) if modules are not lazy coded ( not assuming 44100 SR) all should work perfectly on diverse sample rate, becuase std freq spec it's allways correct respect to runnig sample rate

grease · 28 May 2005

unkargherth wrote: grease wrote: By the way, if frequency values are set with 0..1 representing 0..Nyquist frequency, then this seems to be dependent on what samplerate is being used. Or does the "Nyquist frequency" in SynthMaker always mean 22050 Hz?

That's what y say 0.. Nyquist and not 0..22050. AFAIK (and tested , using asio at SR 48000 ) if modules are not lazy coded ( not assuming 44100 SR) all should work perfectly on diverse sample rate, becuase std freq spec it's allways correct respect to runnig sample rate

Ah ha, I see - I just tried my simple synth example at different sample rates, and indeed it was detuned. So Nyquist does change when samplerate changes, as you said. So I guess I need to know how to code properly without assuming 44100 samplerate. Is there a module that reads host samplerate? I'll look into it...

grease · 28 May 2005

And I just tried FM3x at different samplerates, and it seems to only work at 44100, or else it's detuned. Actually...I think I might start a new thread on this, since I'm a newbie and it might be a newbie kind of question that would help other newbies too...

unkargherth · 28 May 2005

Yes. there's a such module: VST>>Timing info

grease · 28 May 2005

unkargherth wrote: Yes. there's a such module: VST>>Timing info

we posted at the same time! OK..I'm going to start a new thread on this if you don't mind...

jorge · 28 May 2005

grease wrote: And I just tried FM3x at different samplerates, and it seems to only work at 44100, or else it's detuned.

I should warn that although the mechanism to support different sample rates is already implemented, it hasnt been high priority at all and it might not work, we never used it or even test it. Whatever you find out will be very useful.