Fathom Synth Development Thread

[Distorted Horizon]

Or update the current LFO's to support audio rate modulation.

I would love to see Fathom support scales (major / minor / etc) when modulating e.g. pitch/cutoff from a modulator (Seq. / LFO).
Arcsyn supports that feature, and it's awesome.

Yes Maybe a selectable custom scale like in Reason scale player.

[FathomSynth]

I think you can import any scale you want since Fathom now has Microtune.

Sinthoid, RPH, Several people have asked for an audio rate modulation. I'll try to bump that up in the priority list.

[perpetual3]

I think you can import any scale you want since Fathom now has Microtune.

Sinthoid, RPH, Several people have asked for an audio rate modulation. I'll try to bump that up in the priority list.

I think that should be a very high priority.

[RPH]

I think you can import any scale you want since Fathom now has Microtune.

I think we're talking about different things, am not talking about microtuning the keyboard notes / pitch.

Check out the pic.

Source LFO 1 / 2 and 3 use modifier minor harm to destination OSC 1 / 2 and 3 freq.
It's in a Drone patch I love to use

[FathomSynth]

I was talking to Distorted Horizon in regard to the Microtune feature already being there.

In regard to Audio Rate modulation that is already on the list and is a very high priority but has to be done after the CPU parallel processing improvements.

In terms of your picture, generally if you are modulating pitch and you want minor vs major that means you are modulating by steps.

For Fathom that would generally be an arpeggiator feature. Anything that would require Fathom being aware of whether or not the user wanted a modulation to follow major or minor steps would require the arpeggiator logic, since in Fathom modulators themselves have no clue if the song is in a major or minor key.

In terms of your diagram specifically I can't tell for sure what it is doing but it looks like you are asking for a special quantising filter which would take the continuous signal from an LFO and quantise it into 12 steps for a scale either major or minor.

That would be a lot of effort on my part when the user could simply use a sequencer and choose the desired steps from a total quantization range of 12 steps and the selected steps would determine if it was major or minor. I do this all the time for the presets which sound like an arpeggiator but were actually done with the sequencer. The only tricky part is if some chords in a song are major and others are minor then I have to use 2 Fathom tracks, one with a major sequence and the other with a minor sequence.

I'm probably not understanding you so if you can clarify it a little and it's something not too difficult then I can do it.

Incidentally Audio Rate Modulation is probably the highest priority after CPU, Compressor, and Analog Emulator.

[Distorted Horizon]

I wasn't talking about microtuning either but doesn't matter. Feature mentioned would be fun but not important (at least for me)

[jbraner]

In regard to Audio Rate modulation

They do this in Pigment - but they don't worry about actual major/minor keys and the notes - they just let you quantise to various half steps from the note you play.

In other words, if I play a note - I can quantise to a higher (or lowrr) note in half steps. I get to choose which half steps to use - so if I want a major key its 2 4 5 7 9 11 12.
That will be quantised to a major key based on the root note I played - not a particular major key.

Perhaps that would be easier to implement?
Or perhaps not - in which case please ignore this

[wagtunes]

Another Bug:

Assign mod wheel to cutoff frequency to the oscillator filter and it kills sound completely.

[FathomSynth]

JBraner, RPH, Thanks.

Wagtunes, Thanks, that is incredibly helpful. I'll be releasing 2.32 this weekend ahead of a major add campaign and I'd like to make it completely bug and glitch free, so your bug finds are a huge advantage this week. I'm working on the first one you reported here as we speak.

Also, you may have noticed a few audio glitches when you move bezier points in the waveform or modulator envelope views, or when you move the partials dial in wavedraw or wave table. I'm currently working on a new thread design for 2.32 which should completely eliminate this. 2.32 is going to be the cleanest release yet. It will also have an Intel AVX CPU detector in the settings page in preparations for AVX parallel processing very soon after in version 2.33.

Edit: OK, both problems are fixed. It was caused by Fathom not processing the midi input signal unless the host is playing. Now all midi signals are updated even if the host transport is not playing. Now when you move the signal the modulation will be updated regardless of the play head, and it will also update when the host song point is simply moved.

There is one obscure scenario which it can not handle, if the signal envelope in the host is flat and contains no points and you add a modulation using that midi control signal and you hit play without moving the host play head point, then the modulation will not be updated. The reason is that in this case there is no signal message from the host, so there is nothing Fathom can do about that. However once you change the play head point, or add a point to the signal envelope it will be updated.

RPH, Edit, Sequencer gone bug is now fixed.

[RPH]

Hi, please look at this video I made Everett.
Check out the spectrum in SPAN, there are peaks which should not be there.
When I move my mouse, they stop untill I set a different partial value or OSC type.
Sometimes I also get a random "pop", have not figured out when that happens exactly.

Notes:
- It plays one note at C2
- Asio driver, 44.1kHz
- Reaper 5.97

[FathomSynth]

The spikes after the last partial is caused by the reverse FFT spectrum which is used during mouse movements.

Using the reverse FFT instead of recalculating everything is much faster but is not as high quality and creates a dirty spectrum after the last partial. This is necessary since recalculating all the audio bufferes for the oscillator in high quality mode takes too long during mouse drags over a dial and will cause bad stuttering if it was used. So the fast temporary calculation is used while the mouse is being dragged and the dirty spectrum is the trade off to avoid bad stuttering.

Then once the mouse is released it makes the final max quality buffer calculation.

I can not change that during mouse drags.

You will still get the dirty spectrum during mouse drags, I don't think I can change that until after AVX. The new AVX speed might give me the CPU power to get rid of that temporary calculation, but I'll have to test it and see.

Edit: I tried to recreate the problem exactly as you do in your video and I can not make it happen after I release the mouse no matter what I do. I think in your video during that long period of time where the spectrum spikes are there you are actually still holding the mouse down. 0:30-2:00, Is that correct?

[RPH]

Hmm, I see. That does explain why it dissapears.

No, am not holding mouse down between 0:30 and 2.00.
I used the scrollwheel to change the partial parameter 1 step.
Then I waited, was expecting a "pop" to get rid of the spikes. But that didn't happen this time, before it did but of course not when recording..
The only time I used the mouse was to change settings or point at the spectrum in SPAN.

[FathomSynth]

OK, That makes sense.

If you are using the mouse wheel it will go into fast mode until the mouse exits the dial. So the work should be as follows:

Mouse Drag, Fast Buffers, Mouse Up, Quality Buffers
Mouse Wheel, Fast Buffers, Mouse Exit, Quality Buffers

If you isolate the pop let me know since I'm trying to eliminate all pops and glitches for the next release.

[RPH]

Turning on detune to e.g. a sine OSC adds lots of signal (starting at 1kHz) to the spectrum, probably even aliasing if I see it correctly using some spectrum analysers. It's really soft tho, -108dB is of course inaudible.
It doesn't show in the spectrum analyser in Fathom, probably due to not enough dB range.
What also doesn't show is a fairly big bump in the 20Hz range, can't filter it out with the global EQ and the built in HP filters are not surgical enough.

Added a screenshot, Everett?

C2: https://i.imgur.com/QjwYjUV.jpg
C7: https://i.imgur.com/9QWm0bG.jpg

I seem to get clicks when adding a easy ADSR to my testpatch.
It just contains a basic sine OSC with detune and FM on. No distortion.

[FathomSynth]

Your images have FM enabled and you will always get aliasing with FM especially with fractional ratios.

Below is my spectrum with Fathom pure sin and with detune two channels no chorus.

FathomSin.jpg

FathomSinDetune.jpg

From the spectrum we can see that the noise floor is:

-(168-18) = -150 dB
-150 dB = 20 * Log10(noise)
noise = 3.16e-8 = 0.0000000316
Which is 32 one-billionths of the amplitude.

For the ADSR clicks try playing with Envelope Modulator Smooth Samples.