One Synth Challenge #150: TAL-NoiseMaker (Aiynzahev wins!)

[briefcasemanx]

I just realized I have no idea how the vote tallying works. Is it weighted or averaged somehow so that if someone votes mostly 5's their votes don't have more power than someone who votes mostly 3's?

[Double Tap]

I just realized I have no idea how the vote tallying works. Is it weighted or averaged somehow so that if someone votes mostly 5's their votes don't have more power than someone who votes mostly 3's?

No, because it doesn't really matter if you're generous or parsimonious, since you're giving points to everyone. It would matter if you could only vote for a certain number of people. But your votes are effectively a rough ranking of all participants.

Everybody's vote gets tallied up, and usually the winner is the one who gets the most 5s.Take a look at last month's contest - there's a post that Richard puts up detailing the winners and there's a link to a spreadsheet of the results. You can see who voted for whom and there's a 'generosity index' which indicates whether a person has given out lots of fives or lots of twos. It's probably easier to understand if you look at the numbers.

[z.prime]

is it against the rules to use Cymatics samples? B/c i mean they're fire bro. that's what their emails say. i hear that is all @the never scene and @z' use.

Yeah, it's fine unless you get checked bruh, just say you rendered the synth. I usually do a collab. I start with Fiverr. I get a track back and then sometimes think I might change it up. But I never really know how to, and I can't really make the same sounds they used, so I just write a program that takes the song and automates volume on any synth so it just outputs the waveform. It's worked out pretty well, bruh. Let's have another threesome collab with my Fiverr bros this month!

[Aro]

I dont know if I fully understand what Peter H did, but I do appreciate the discussion surrounding it. Its upgrading my higher thinking abilities for sure. Still trying to grasp at how envelopes can be like samples. Its a slippery thought for me.

[briefcasemanx]

I dont know if I fully understand what Peter H did, but I do appreciate the discussion surrounding it. Its upgrading my higher thinking abilities for sure. Still trying to grasp at how envelopes can be like samples. Its a slippery thought for me.

I'm confused a bit by the technical details as well so I have questions. It seems like he's doing amplitude modulation: https://en.m.wikipedia.org/wiki/Amplitude_modulation
You can see in the first animation how one signal can be encoded onto another signal called a carrier. I think usually for like AM radio they would eliminate the carrier signal before it hits your speakers, which should be easy since the carrier signal is a known frequency. Eliminating the carrier is called demodulation.

But there is no demodulation done when using amplitude modulation in the MSEG, so would the end result be a highly distorted sample with a DC offset?

I'm assuming DC offset because you can't have negative amplitude in the MSEG. So the amplitude of the waveform you're sampling must have all positive values. If you sample a sine wave in the MSEG with a ton of points, what is normally the zero crossing for a sine will be at half of max amplitude of the wave. And the negative "trough" of the sine wave would be at a volume of zero or higher. All values of the sine wave would need to be positive values inside the MSEG, which is equivalent to a sine wave with a DC offset.

I'm assuming there's be a lot of distortion because the carrier wave is not eliminated (demodulated) from the signal and would be audible. And also because depending on how the MSEG is coded and it's internal resolution there could be other distortion like aliasing I'm thinking, because with all those points it's kind of close to a raw digital signal of sample points, but I there may be no anti-aliasing filter in the MSEG since 10,000 points is way, way beyond it's design specification.

I wonder if the considerations for the frequency of the carrier signal might be different in this situation than a normal AM implementation with carrier demodulation? What would happen if the carrier signal is outside the audible range.

So in my confused and limited understanding it **seems** like it should produce a heavily distorted sample probably with DC bias if it's using amplitude modulation, but I have literally no clue what I'm talking about so please someone correct me if none of this makes sense.

[philnsicab]

About the NoiseMaker plugin : there is a filter called Cl LP24. Do you know what type of filter it is? What Cl stand for?

'classic' moog-style 4-pole ?

On second thought: maybe 'classic' Roland Juno 4-pole is more likely.

Do you have a source or a hint for that ? Maybe a statement from the developers? I couldnt find the filter in the manual.

[VCA-089]

Voting done!

[becseigy]

Voted.

[neverbeeninariot]

Voted.
Tough choice though, there are lots of great entries. Good luck everyone

[IrionDaRonin]

Voted. Good luck everyone!

Have a nice day

[] Peter:H []

I dont know if I fully understand what Peter H did, but I do appreciate the discussion surrounding it. Its upgrading my higher thinking abilities for sure. Still trying to grasp at how envelopes can be like samples. Its a slippery thought for me.

I'm confused a bit by the technical details as well so I have questions. It seems like he's doing amplitude modulation: https://en.m.wikipedia.org/wiki/Amplitude_modulation
You can see in the first animation how one signal can be encoded onto another signal called a carrier. I think usually for like AM radio they would eliminate the carrier signal before it hits your speakers, which should be easy since the carrier signal is a known frequency. Eliminating the carrier is called demodulation.

But there is no demodulation done when using amplitude modulation in the MSEG, so would the end result be a highly distorted sample with a DC offset?

I'm assuming DC offset because you can't have negative amplitude in the MSEG. So the amplitude of the waveform you're sampling must have all positive values. If you sample a sine wave in the MSEG with a ton of points, what is normally the zero crossing for a sine will be at half of max amplitude of the wave. And the negative "trough" of the sine wave would be at a volume of zero or higher. All values of the sine wave would need to be positive values inside the MSEG, which is equivalent to a sine wave with a DC offset.

I'm assuming there's be a lot of distortion because the carrier wave is not eliminated (demodulated) from the signal and would be audible. And also because depending on how the MSEG is coded and it's internal resolution there could be other distortion like aliasing I'm thinking, because with all those points it's kind of close to a raw digital signal of sample points, but I there may be no anti-aliasing filter in the MSEG since 10,000 points is way, way beyond it's design specification.

I wonder if the considerations for the frequency of the carrier signal might be different in this situation than a normal AM implementation with carrier demodulation? What would happen if the carrier signal is outside the audible range.

So in my confused and limited understanding it **seems** like it should produce a heavily distorted sample probably with DC bias if it's using amplitude modulation, but I have literally no clue what I'm talking about so please someone correct me if none of this makes sense.

Okay, quite a few points here.
Yes, I've worked with amplidute modulation. But not in a sence of transmitting radio waves like a radio station ... because you have asked for demodulator.
By the way - Amplitude modulation is different to PCM (pulse code modulation) which is used in what we know as wav file aka a "sample" in a traditional sense.
Amplitude modulation can work in a "slow common setup", like you use a sine wave with 80 hz and modulate volume (i.e. amplitude) with a 2 hz lfo. We all know what happens.
But what happens when you increase the frequency in the LFO? You can do this in NoiseMaker - Just go up to 160 Hz and modulate Volume or Pan. Now new fequency content is emerging.
What I do: I use a very "detailed AM" and run it in x32 mode, that means more on the fast side than on the slow side.
Side note: In contrast to what the music industry knows as sampling the MSEG can only run in one of four speeds, that means in contrast to a real sampler a MESG can never ever reproduce every pitch AND it's completely fixed. Once you set the mode (x2, x4, x32) then it's set and done.
When I now let this fast running detailed AM be combined with a good selection of OSC - which really needs sme work as well. not all OSC setting work - then we get close to an effect of PCM. And thats exactly why it kind of works.
But you are right - It's distorted. a.) the MSEG is only positive values. so it cannot model the positive and negative half waves of a real PCM sample exactly. And b.) there's a "transformation in time frames" because x32 in song tempo is another time reference frame then 11.050 PCM samples per second.

Anyways I love these kinds of experiments. Because I know a little how sound works, how AM can be used to "only mimic" pcm to some extent. So I'll just give it a try and giggle like a mad scientist in it's cave laboratory if the idea somehow works.

If you still say this is "sampling" in a traditional music industry sence ... then go ahead ... I will not stop anybody doing so.
But the summary is:
1.) It's AM not PCM (Samples are based on PCM ... kind of common sence)
2.) It can reproduce a "distorted" copy of things to some extent but is very very limited.
3.) It cannot transpose over the full key range
4.) The MSEG has only positive values in the range [0,1]
5.) It cannot reproduce the source 1:1. And that's by the way where I and my programm was wrong. I forgot to transform the formula into the correct time reference frame because x32 in the MSEG is relative to the song tempo... and if you want to have a frequency of 170 hz in a x32 MSEG on song tempo of 154 bpm means you would have to use another base frequency ... my bad ... it still got some nice sound

Don't know whether this makes sence for you. I've had lot's of fun doing experiments and software for this challenge. And I will not stop doing explorative things in the future... Only thing I stop - I will better think before or even better not share anything anymore. It's not worth the hazzle.

] Peter:H [

[z.prime]

1.) It's AM not PCM (Samples are based on PCM ... kind of common sence)
2.) It can reproduce a "distorted" copy of things to some extent but is very very limited.
3.) It cannot transpose over the full key range
4.) It cannot reproduce the source 1:1. And that's by the way where I and my programm was wrong. I forgot to transform the formula into the correct time reference frame because x32 in the MSEG is relative to the song tempo... and if you want to have a frequency of 170 hz in a x32 MSEG on song tempo of 154 bpm means you would have to use another base frequency ... my bad ... it still got some nice sound

Don't know whether this makes sence for you.

Oh god. Shut up. Posting all this in an effort to confuse people and conflate what you are doing. PCM vs AM is a strawman. You can use AM on a DC offset to achieve exactly the same thing: sample playback.

1) You are effectively using sample playback
2) You are a condescending jerk
3) What you've done is actually very impressive, just not allowed

EDIT: and also: learn to spell.

[Guenon]

If you still say this is "sampling" in a traditional music industry sence ... then go ahead ... I will not stop anybody doing so.

The crux of the matter is, did the end result need an audio recording, at any point of the process, originating from somewhere outside of the synth, containing a spoken word, in order for the speech patch to be able to reproduce that word in the first place.

To reiterate, if you were locked up in a room with a DAW and just Noisemaker installed, no microphone, no internet access, no audio clips - would you still make a track that speaks that word in that manner or not , or is having access to an outside audio recording in fact necessary for it to sound the way it does.

Ideas like this are cool. Being mad in this manner is cool. Actually realizing an idea like this in practice is cool. That isn't the issue. In any case, as a programmer, I personally view the end result of your experiment as the expected outcome, given Noisemaker's feature set and being able to write enough MSEG points programmatically into a valid .noisemakerpreset file, and I wouldn't do it in the context of this challenge, as it does depend on outside audio, to sound the way it does.

Again, actually doing it is cool (shaping reality through programming is cool), yet you know it wouldn't sound the way it did, if you were given the task to sit in a room and do a track in the manner described above - or would it? If yes, that would be the... unexpected outcome, and from my perspective, in this context, more interesting.

Only thing I stop - I will better think before or even better not share anything anymore. It's not worth the hazzle.

You imported and played a recorded spoken word, originating from outside of the synth itself, in the one synth challenge. This doesn't mean the project itself wasn't cool, it doesn't mean you need to stop doing stuff like that, it doesn't mean you need to be melodramatic about it , it's just what it is: in this context, bringing in outside audio like that isn't the best move for an actual entry.

[z.prime]

... did the end result need an audio recording, at any point of the process, originating from somewhere outside of the synth...

Small correction: not an audio recording per-se, but something that contains or has created sample points over time. Again, definition of sampling.

And, all this effort and drama still didn't yield the best track of the competition.

[Guenon]

Small correction: not an audio recording per-se, but something that contains or has created sample points over time. Again, definition of sampling.

I'm naturally on the same page with you on this - otherwise one could also synthesize arbitrary audio with any custom tool and just import data straight from that; however, note that my intention in that passage was just to address the specific spoken word patch Peter used, and when saying "audio recording" I meant that particular speech snippet the patch utters, which I assume was originally an actual audio-ass fully-audioded audio-audio recording.