I tried to keep this one short, but I do have a few additional notes: 1. In my experience, the slower a Shepard tone is, the more convincing it is. And you'll have to slow down the LFOs even more when covering a larger octave range. 2. You need a drone/pad with a static pitch when using samplers, but that's not the case when doing granular Shepard tones! Any sample will do - but you'll probably have to set a fixed grain playback position. 3. Ideally, you want the different voices to be tuned an octave apart from each other so that the human brain is tricked into thinking the upper voices are harmonics of the lowest voice. When using effects such as flangers, comb filters, etc, it's not always easy (or possible) to tune the cutoff frequencies to actual notes, and to have the voices apart by an octave. That is the case in this video (except with the resonators, which are tuned to precise pitches). It's fine, you can still get a decent effect out of it. The sample is 'in tune' with itself and the comb filtering affects its partials without changing their harmonic relationship too much. You could probably get perfectly tuned phasers, flangers and comb filters in a virtual (or physical) Eurorack environment, or by building custom effects in a framework such as JUCE, PureData, Reaktor or Max. (Edit: it used to be possible with the plugins Endless Series by Oli Larkin and Möbius Filter by iZotope, but I think both are discontinued). 4. If you want to print your Shepard tone to audio to make a looped sample out of it (to implement in a game, for example), you don't have to print a whole period of the LFOs to get a perfect loop. You just have to print a fraction of the cycle that's the inverse of the number of voices - for example: a third of the cycle when using three voices. That's because that's the time each voice takes to get to the initial pitch of the neighboring voice. Give yourself a bit of 'pre-roll' and 'post-roll' if you're processing the sound through time-based effects (delay, reverb, etc). 5. You can experiment with loading a different sample for each voice. The effect will be less seamless, but you can achieve more interesting movements (kind of similar to the sounds I made in this video when modulating the grain playback position somewhat randomly).
Amazing as always! Could you please explain the part where you say it's a range of 3 octave so you divide by 2 ? I am not sure to understand here and I feel it's an important part also for adding extra voices! Thank you! :)
Hey thanks James! Are you talking about the part at 05:28? So: the pitch LFOs are set to bipolar mode. They output modulation values ranging from -1 to 1 (instead of 0 to 1, in unipolar mode). So a bipolar LFO covers twice as much range as a unipolar LFO! In the first example (2 voices covering 3 octaves), by setting a pitch modulation value of 12 semitones (= 1 octave), we cover 12 * 2 = 24 semitones = 2 octaves. So in the second example (3 voices covering 3 octaves), we cover 18 * 2 = 36 semitones = 3 octaves. I hope this clarifies it. Let me know if you have further questions 😄
genius idea and the sound is fascinating, too. i have an idea why it might sound less convincing on higher speeds though. I noticed that your gain shapes are just triangles, but the lowest part of the triangle doesn't seem to be super quiet. maybe it would be better to use a sharper shape on it that resembles decibel scaled gain parameters a bit more. a softer transition from silence to sound likely enables you to enjoy the higher speeds. i could imagine a patch like this sounding really cool on various temposync rates
Thanks man! ❤ About the volume LFOs: I experimented with A LOT of different shapes/window functions. Hanning, cosine, exponential, logarithmic... For context: I was looking for practical ways to make Shepard tones before I even got to use Phase Plant ahah. I made Shepard tones in PureData, Voltage Modular, with Python scripts... All that to say: I experimented a lot around this idea. (As it turns out, Phase Plant is the perfect tool for that!) And while the volume envelope shape does have an impact on the sound, at higher rates, it's really just that the human ear is better at catching the coming & going of sounds that fade in & out too quickly. But I do encourage you to try! Maybe you'll have cooler results than me. Please post the results and ping me! I'm always interested to listen to other people's experiments 😄
could you please please please make a longer version of the hypnotoad avalible as a donwload, i really want to make an infinity loop video out of it for discord ^^
![Pt 2 - Sampling: risers [Sound design workshop]](http://i.ytimg.com/vi/UMpLPzC4Ck0/mqdefault.jpg)
![Pt 2 - Sampling: risers [Sound design workshop]](/img/tr.png)
I tried to keep this one short, but I do have a few additional notes:
1. In my experience, the slower a Shepard tone is, the more convincing it is. And you'll have to slow down the LFOs even more when covering a larger octave range.
2. You need a drone/pad with a static pitch when using samplers, but that's not the case when doing granular Shepard tones! Any sample will do - but you'll probably have to set a fixed grain playback position.
3. Ideally, you want the different voices to be tuned an octave apart from each other so that the human brain is tricked into thinking the upper voices are harmonics of the lowest voice. When using effects such as flangers, comb filters, etc, it's not always easy (or possible) to tune the cutoff frequencies to actual notes, and to have the voices apart by an octave. That is the case in this video (except with the resonators, which are tuned to precise pitches). It's fine, you can still get a decent effect out of it. The sample is 'in tune' with itself and the comb filtering affects its partials without changing their harmonic relationship too much. You could probably get perfectly tuned phasers, flangers and comb filters in a virtual (or physical) Eurorack environment, or by building custom effects in a framework such as JUCE, PureData, Reaktor or Max. (Edit: it used to be possible with the plugins Endless Series by Oli Larkin and Möbius Filter by iZotope, but I think both are discontinued).
4. If you want to print your Shepard tone to audio to make a looped sample out of it (to implement in a game, for example), you don't have to print a whole period of the LFOs to get a perfect loop. You just have to print a fraction of the cycle that's the inverse of the number of voices - for example: a third of the cycle when using three voices. That's because that's the time each voice takes to get to the initial pitch of the neighboring voice. Give yourself a bit of 'pre-roll' and 'post-roll' if you're processing the sound through time-based effects (delay, reverb, etc).
5. You can experiment with loading a different sample for each voice. The effect will be less seamless, but you can achieve more interesting movements (kind of similar to the sounds I made in this video when modulating the grain playback position somewhat randomly).
This patch + granular sources = perfect psytrance risers
Yes sir! Do it!!
Cool! Merci 👍
This reminded me of Elite Dangerous sound design for accelerating spaceship drives. Fucking sick
That's a huge compliment! That game sounds insanely good
I have found your tutorials to be extremely helpful and informative. Please continue to create more videos!
Thanks for the kind words man, it's really encouraging to read 💚
More videos are definitely coming!
The bass music builds from this technique are going to be insane
Do it!!!
I discovered this channel yesterday, super nice stuff!
Hey thanks man! I appreciate the kind words, and the subscription!
Thanks for these videos!!! Always so inspirational!!!
Thank you! 🙏
INSANE work as usual!
Thanks!! Glad you enjoyed it!
Great video.
Was able to port this tech to the Korg Modwave hardware in about 15 minutes minutes.
Awesome! I'd love to hear the result
Amazing as always!
Could you please explain the part where you say it's a range of 3 octave so you divide by 2 ? I am not sure to understand here and I feel it's an important part also for adding extra voices!
Thank you! :)
Hey thanks James!
Are you talking about the part at 05:28?
So: the pitch LFOs are set to bipolar mode. They output modulation values ranging from -1 to 1 (instead of 0 to 1, in unipolar mode). So a bipolar LFO covers twice as much range as a unipolar LFO!
In the first example (2 voices covering 3 octaves), by setting a pitch modulation value of 12 semitones (= 1 octave), we cover 12 * 2 = 24 semitones = 2 octaves. So in the second example (3 voices covering 3 octaves), we cover 18 * 2 = 36 semitones = 3 octaves.
I hope this clarifies it. Let me know if you have further questions 😄
wuzzah, great idea and explained really well!
Hey thanks! 😄
genius idea and the sound is fascinating, too. i have an idea why it might sound less convincing on higher speeds though. I noticed that your gain shapes are just triangles, but the lowest part of the triangle doesn't seem to be super quiet. maybe it would be better to use a sharper shape on it that resembles decibel scaled gain parameters a bit more. a softer transition from silence to sound likely enables you to enjoy the higher speeds. i could imagine a patch like this sounding really cool on various temposync rates
Thanks man! ❤
About the volume LFOs: I experimented with A LOT of different shapes/window functions. Hanning, cosine, exponential, logarithmic...
For context: I was looking for practical ways to make Shepard tones before I even got to use Phase Plant ahah. I made Shepard tones in PureData, Voltage Modular, with Python scripts... All that to say: I experimented a lot around this idea. (As it turns out, Phase Plant is the perfect tool for that!)
And while the volume envelope shape does have an impact on the sound, at higher rates, it's really just that the human ear is better at catching the coming & going of sounds that fade in & out too quickly.
But I do encourage you to try! Maybe you'll have cooler results than me. Please post the results and ping me! I'm always interested to listen to other people's experiments 😄
It’s not easy to get this to sound right. Even if everyone knows the theory. This is by far the best sonic example!
Hey, thanks for the kind words man!
Make a forever THX!
They sound terrible and I love them! :) Great video, awesome sounds.
Ahahah thanks! 😄
Im gonna die of infinite build up 😢❤
Better than dying of infinite let down! 😄
@@HoMinhAudio infinite build up then it ends and no drop 🔥
Where are you from ??
could you please please please make a longer version of the hypnotoad avalible as a donwload,
i really want to make an infinity loop video out of it for discord ^^
I'll get back to you on that!