Loopy Pro: Create music, your way.
What is Loopy Pro? — Loopy Pro is a powerful, flexible, and intuitive live looper, sampler, clip launcher and DAW for iPhone and iPad. At its core, it allows you to record and layer sounds in real-time to create complex musical arrangements. But it doesn’t stop there—Loopy Pro offers advanced tools to customize your workflow, build dynamic performance setups, and create a seamless connection between instruments, effects, and external gear.
Use it for live looping, sequencing, arranging, mixing, and much more. Whether you're a live performer, a producer, or just experimenting with sound, Loopy Pro helps you take control of your creative process.
Download on the App StoreLoopy Pro is your all-in-one musical toolkit. Try it for free today.
jAmp Convolve by Jan Maes (Released)
https://apps.apple.com/app/id6758905873
jAmp Convolve is a professional audio convolution plugin (AUv3) for iOS and macOS, using advanced neural network‑powered impulse response generation to deliver realistic guitar cabinet, bass cabinet, and reverb simulations — ideal for musicians, producers, and sound designers. Unlike traditional convolution plugins, it does not require loading external impulse response files. Instead, jAmp Convolve’s AI engine dynamically shapes the sound in real time, letting you explore and fine‑tune tones without the hassle of managing IR libraries.
Details:
Universal: Yes
Minimum OS version: 14.0
Comments
That thing is really good?
@janm31415 ,are all files speakers only?
Thanks.
Hi Flo,
The Guitar and Bass tabs generate IRs (Impulse Responses) that simulate cabinets (I assume this corresponds with "speakers only" if I understand the question correctly).
The Reverb tab will generate IRs that represent a room or hall.
If you want to use a speaker IR combined with a room IR, you will have to use two instances of jAmp Convolve in your DAW.
@janm31415 ,thanks a lot for the answer.
Yay, finally something for guitarists
@janm31415:
Hi Filo,
The red and green dot together control 4 parameters, i.e. each dot has a (x,y) coordinate. These 4 parameters are the input to a neural network that I trained on IRs. Depending on the value of these 4 parameters (or the position of the red and green dot), you get a different impulse response from the neural network. There is however no other sensible or logical mathematical idea behind the location of the red/green dot. I mean by that, that it's not like you can say that going to the right with a dot location will make the sound sharper while going to the left makes it softer. The neural network is a bit unpredictable in that way, except for the fact that you always get a valid impulse response. Small movements with the dots will slightly alter the impulse response, but the general idea is that you have to try moving the dots until you like what you hear. Save your favorite spots as a preset, or start from the existing presets and experiment.
The "export" functionality is only for the presets, so there is no real export of the impulse response wavs. Not providing a real export of the impulse response wavs is kind of a deliberate decision, as I'd like people to use rather my AUv3 instead of an IR wav that was generated with it. So I understand that having a real export has value, but (from a commercial point of view) I'm intentionally keeping it limited to presets in order to encourage using the AUv3 itself rather than distributing standalone IR WAVs.
Jan, thank you very much for the detailed explanation!
I assume that by "AI" all you mean is an embedded and trained net, and not something that actually phones out to an externally hosted complex model?
Yes, that is correct.
Wow this is great @janm31415
Thanks!
Groovy. May I ask if you'd be willing to share some generic technical details? Number of layers, general scale of parameters, size /breadth of training set, that kind of thing.
Sure, the system is based on a convolutional autoencoder with a very compact latent (4 dimensions) representation, trained on 2D time–frequency representations (spectrograms).
Multiple convolutional stages (6 convolutional encoders and 6 convolutional decoders) are used, and as a result the total parameter count is in the low‑to‑mid millions, not tens or hundreds of millions. This is just large enough to model complex IR structure, without the need to have some state of the art equipment.
Training is done on 5000 IRs for the guitar, 1500 IRs for the bass, and 300 IRs for the reverb.
The 4-dimension latent space choice makes this very interesting to me. One problem I've always had with typical IR loaders is the choice of so many IRs out there. If you can find one that matches your guitar perfectly - great! But that is not so easy. I like the idea of dialing in sound in a cleaner, simpler way == morphing between tonal characteristics (rather than having to load different files and testing them out individually.)
Might lose some hyper-realism of using custom IRs, but I think more sound control is worth it.
I'm going to get this.
@tubespace - that is an excellent point.