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.
Comments
Maybe "guess" is the wrong word. It regularly, and I mean regularly, confidently gives answers when clearly it has no clue what it's talking about. Yes, I think hallucinating, is a better term. It simply doesn't seem to know when it doesn't know something. Instead of saying that it doesn't know something, it puts out answers that are completely wrong.
I'm not making this up. It happens regularly in just about every session I've ever conducted. I need to be the one to filter out what I ask to avoid getting a load of BS back.
Sorry. Off topic. I'll leave off.
ChatGPT is essentially a predictive text engine trained on an enormously database (corpus) with a sophisticated machine learning algorithm. It doesn’t reason through solutions. If the corpus has an overwhelming amount of text related to a subject that is correct, it will generally provide text that is both well-constructed.
But if the corpus has wrong or ambiguous text in it, it will spit out well formed but wrong or ambiguous text.
It gets lots of stuff wrong. And sometimes what it predicts is an accurate prediction of a wrong thing someone might say.
How can “guess” be an inappropriate word for a process based on probability?
This is something I have been struggling with also, there is no problem using prog change in Nave, but nothing I can do will make it work in Microwave.
I have contacted Waldorf using App Support and a ticket has been raised, so hopefully I will get an answer and pass it on 👍
Actually whether or not Large Language Models “guess” is a complicated subject. Generally speaking, LLMs are non-deterministic. “Temperature” is the parameter used to control sampling probability/“greediness”. One can set it within a reasonable range, but rarely/never to 0- doing so will often break results: e.g. getting the same words too often following certain words, stuck in loops, or not providing results that are specific/specialised enough. Randomness is not just used to control creativity. An element of non-determinism is also key to general performance.
Even ignoring the element of run-time determinism- consider the meaning of the word “guess” as people use it. If I don’t “know” the answer to something I’m asked, but I give an answer regardless, based on what I think is most probable- most people would call that “guessing”.
LLMs are just gigantic language probability machines, and don’t “understand” whether they “know” anything. They only work with probabilities- “guessing”. If you prompt a model for a list of musicians who used a specific piece of equipment, it might have an unreasonable level of confidence as it explores latent space between associated subjects. Maybe the Beatles never used a Waldorf Microwave- but perhaps there were other strong associations in the training data- articles where both topics are mentioned frequently, or closely associated bands who did use a Microwave. It’s very hard to reverse-engineer the associations encoded in the models, because they’re so gigantic. The latent space is fundamental to LLM magic, but we don’t know how to sample it without entirely eliminating “hallucinations”. One could argue the hallucinations are actually fundamental to how they work.
TLDR- I would phrase it that LLMs are always guessing. Sometimes they’re very reasonable guesses! Unfortunately we sometimes can’t tell. If we programmed them to stop guessing, they wouldn’t work at all.
Can we pls get back to topic? Thanks.
An interesting read, thank you!
I just disagree that “LLMs are non-deterministic”, they totally are deterministic as they follow an algorithm. It’s just the absurd amount of data that is impossible for us to comprehend, but they still follow a reproducible number of steps, that if we could trace (and tools for that exist and are being improved) each and every billionth step we’d arrive at the exact same result.
Adding random factors to it doesn’t make it non-deterministic either, cause there isn’t a true random generator in computing, only pseudo random, as in if you start from the same seed you get the same results (why the seed is taken from a variety of sources available to the cpu, like I dunno room temperature, but even then it’s still deterministic as the computer starting from the exact same initial conditions, would produce the exact same results)
/being pedantic
My main motivation for buying this would be to encourage Waldorf to revisit Nave.
☝️☝️☝️☝️
I have a question: I would like to run through the wavetables (a classical wavetable scan) using the wave envelope. The envelope should start at note on and run through. And stop at the last wavetable, disregarding note off.
With the PPG wave 2.2 this is done instantly, but I’m not familiar with the successor, the MicroWave. The MicroWave has a complex envelope, but I cannot see a way to create a simple envelope.
There's no need to use all the envelope steps, set the Key Off Point to 1 and adjust Time 1 to ramp up from start to end and if you want it to sweep back on release adjust time to and set level to 0. After that set the wave Env to +60 (to avoid the last few basic waves).
I can’t check at the moment, but isn’t there a preset that does what you want? Presuming you mean a basic wave table scan. I may have misunderstood your question though.
i stumbled over something interesting in the german microwave thread on sequencer.de (Rolf is a bit active there). Basically saying „if someone with a (Waldorf) Wave is bored, message me „
Well, that can mean something or nothing. 😎
@Samu and @Fruitbat1919, would it be possible to post a screenshot of the wave envelope, or point to a preset that does wavescan as I would like?
I just work with the first preset “Waveride”, and fail to accomplish this simple edit. Key Off Point is set to 1, that was one of the first things I’ve edited. Adjusting Time and Level do not lead to the desired result.
Maybe I should ask the question differently: is it even possible to start and run the wave envelope at Note on without holding the key?
one can hear the wavescan sounds in this song:
This was a PPG wave 2.0. With its successor, the wave 2.2, it is just one or two edits, and the wavetable runs through to the end or earlier, depending on the env>wave knob.
Did you try enabling loop for the wave envelope?
This way when the key off point is set to 1 it will start looping after you release the key letting the wavetable scan during note release.
(Sped up the envelope times a bit to 50).
Sometimes I cannot choose presets from the dropdown menu, only the back or next buttons work (Apematrix). Is this an issue on my side?
sometimes I cannot change presets with those back or next buttons. might be related.
I had the same. Seems a bit buggy.
Always loved the PPG Wave sounds in this intro:
Has anyone tinkered with the “stepped” wave mode? It doesn’t seem to have any effect. What I would expect to hear is that weird gurgle, as the scanner moves through the table without interpolation, but I don’t think it’s there at all. Possibly unimplemented?
This is a Great instrument!
I was not going to get any new synths but this one is amazing and i had to grab it!
Highly recommend!
@nuno_agogo @Gavinski I submitted a bug report (and additional wavetables request
) on the Waldorf website, we’ll see if it will land on the right desk.
This site also mentions the Roland System 700 being used on Enjoy The Silence: https://dmlive.wiki/wiki/Enjoy_The_Silence
Thnx, I am yet to ever get any reply from them
Does feel like it doesn't really step thru the table when using a slow LFO or Envelope but there's slight difference, there's a lot less 'clicks' when the 'Smooth' option is selected, more like it doesn't change the wave 'mid-phase' but rather waits until it's finished it's cycle while 'Steppec'(Spelling error?) reduces the amount of clicks?!
The whole album" A Broken Frame", has the PPG 2.0 all over it.