Oscillators in VSTi?

[fluffy_little_something]

We also could say that every digital oscillator produces numbers. They can be precomputed (and read from a lookup table) or computed in realtime. We hear sound because of digital-to-analog conversion (DAC).

So, basically the computed "signal" stands for the amount and way electrical current is to be pumped through speakers

[lobanov]

There are no any really significant difference between the digital oscillator in a hardware synth and the one in a software synth. The hardware uses a dedicated processor, the software uses recources of the computer.

Sampler is a sort of a digital oscillator. But it's different from an oscillator in the ordinary, usual sense by the way it reads and reproduces data. The oscillator based on the lookup table reprodices it looped, the sampler reproduces the data linearly, from the beginning to the end (in a basic case). But the general principle is the same: read numbers and send them to DAC. There are additional tricks to prevent the aliasing but it's a different story.

[Dasheesh]

^^^ thus every oscillator sounds the same and there is no validity to “analog sounds alive”.

80% of people are still doing it wrong IMO. It’s a new world and they haven’t adapted. That includes most devs.

[teacue]

These are a lot of interesting thoughts

In particular this:

A software oscillator is an algorithm, one that digitally generates a signal to a specified frequency. That 'specified frequency' bit pretty much means that there's always some sort of algorithm in there, even if all the oscillator does in terms of the waveform is look up a precalculated single-cycle waveform data source and resample that to the required frequency.

More sophisticated oscillators can be based on more complex algorithms, 'up to' entirely algorithmic calculation of the waveform with no usage of a data source. Some synthesis methods, though, (eg scanned multi-cycle wavetables,) intrinsically rely on data sources.

'Data source' in this context can be 'sample' or 'wavetable' or whatever name for a stored waveform you want to use.

and this:

We also could say that every digital oscillator produces numbers. They can be precomputed (and read from a lookup table) or computed in realtime. We hear sound because of digital-to-analog conversion (DAC).

help me a lot to better understand.
Thanks for these explanations

Now in term of audio results or audio quality what is the difference for an oscillator being either:
1. a computed waveform
2. a precomputed waveform
3. or a sample of a single cycle waveform?

[Dasheesh]

^^^ it’s false. Electricity IS alive. It produces different results every time it cycles. That’s too much cpu for the edm crowd tho, and it really doesn’t matter in today’s radio production. It’s ALL digital so no body cares or hears it.

[Dasheesh]

This all really does go back to the music industry. They kept trying to change up the format to keep control, eventually they reached digital format and didn’t realize the impact it would have. Looking back on it... computer sound probably would have arrived anyway. Here’s the thing. Digital DOESNT SOUND BAD. It sounds great! It is different and it does need to be used differently because sound reproduction and use grew with reproduction techniques. You have to hear a good musical sound. And you hav to use a good musical sound. Both of those are lost on 80% who are imitating what they know. Not making a new standard.

[whyterabbyt]

Sure a sine sounds like a sine, but those are just words. Just like blue is just a word, but it has a very specific physical background. With sine it is the extreme simplicity I think, no harmonics.

'sine' has a far more specific 'physical background'.

[whyterabbyt]

help me a lot to better understand.
Thanks for these explanations

Now in term of audio results or audio quality what is the difference for an oscillator being either:
1. a computed waveform
2. a precomputed waveform
3. or a sample of a single cycle waveform?

There's no intrinsic quality to any of those things. Any given instance of one of them could be better or worse than any given instance of a different one.

[teacue]

help me a lot to better understand.
Thanks for these explanations

Now in term of audio results or audio quality what is the difference for an oscillator being either:
1. a computed waveform
2. a precomputed waveform
3. or a sample of a single cycle waveform?

There's no intrinsic quality to any of those things. Any given instance of one of them could be better or worse than any given instance of a different one.

OK.
Thanks

[Dasheesh]

help me a lot to better understand.
Thanks for these explanations

Now in term of audio results or audio quality what is the difference for an oscillator being either:
1. a computed waveform
2. a precomputed waveform
3. or a sample of a single cycle waveform?

There's no intrinsic quality to any of those things. Any given instance of one of them could be better or worse than any given instance of a different one.

Actually.. correct. It’s because everything is produced and replayed digitally. That moots anything live unless it’s being played live... even then, when is an audience listener going to distinguish between analog and digital? I probably wouldn’t and I’m looking for it. Only the player knows the difference now. That’s why my whole argument is .. give up your pride. Sell your self. Nobody cares anymore. And yes, it’s extremely difficult to do.

[whyterabbyt]

^^^ it’s false. Electricity IS alive. It produces different results every time it cycles. That’s too much cpu for the edm crowd tho, and it really doesn’t matter in today’s radio production. It’s ALL digital so no body cares or hears it.

Nobody hears anything digital. Its all been converted into electricity, then magnetism, then kinetic energy before it reaches your ears. A DAC produces electricity, and even the crudest, simplest, lowest-CPU algorithm, post-DAC, produces a different result every time it cycles.

[whyterabbyt]

Actually.. correct. It’s because everything is produced and replayed digitally.

No, that's not why its correct. Its correct because any of the given methods can be implemented with any arbitrary level of quality.

That moots anything live unless it’s being played live... even then, when is an audience listener going to distinguish between analog and digital? I probably wouldn’t and I’m looking for it. Only the player knows the difference now. That’s why my whole argument is .. give up your pride. Sell your self. Nobody cares anymore. And yes, it’s extremely difficult to do.

'Playing live' or distinguishing analog from digital has zero relevance to the discussion.

[Dasheesh]

^^^ it’s false. Electricity IS alive. It produces different results every time it cycles. That’s too much cpu for the edm crowd tho, and it really doesn’t matter in today’s radio production. It’s ALL digital so no body cares or hears it.

Nobody hears anything digital. Its all been converted into electricity, then magnetism, then kinetic energy before it reaches your ears. A DAC produces electricity, and even the crudest, simplest, lowest-CPU algorithm, post-DAC, produces a different result every time it cycles.

Well, perhaps you are correct in listening to playback. It’s reproduced through a magnet. It’s just not how it started and not how the last generation learned it, and it’s hard to come to grips with. We did it because it was fascinating, not because it was expected.

[whyterabbyt]

It’s just not how it started and not how the last generation learned it, and it’s hard to come to grips with. We did it because it was fascinating, not because it was expected.

I have no idea what 'it' you are talking about. 'It' seems to bear no relationship to a basic explanation of digital oscillators.

[chk071]

I always wondered which soft synths calculate the waveforms, and which are sample based. Seems like noone cares really. My understanding was though, that most calculate the waveforms. What confused me a bit though is that the talk is of "wavetables", when it comes to unison on oscillator level.