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
Oh cmon man, lol
Watch the video for a demonstration of how great the quality is . I never knew about “Let’s unmix”.. How’s the quality there? Thanks for bringing it to my attention..
Love the direction they are going.. noticed the ‘Mastering’ thing also.. Maybe I will get to it later..
Oh.. Ableton Link would be nice..
He said second to none…
@RajahP : Yeah I believe Ave . I saw that too. But I forgot about Moises. Maybe it’s not as good but it will probably improve . It seems good right now though .
Moises and some other products actually are built using Spleeter to do the separation. Spleeter is open source and uses the MIT License, which allows anyone to use it with basically no strings attached: https://github.com/deezer/spleeter
Same with Demucs, an open source ai splitter project that comes from Facebook/Meta and is used in, e.g, Audiostrip.co.uk and MVSep.com websites: https://github.com/facebookresearch/demucs
Anybody who wants can get Spleeter and/or Demucs off github for free and use as much as you want. If not the best, these are among the best right now. My guess is that over time the open source ai splitter projects will win out over others, and most of the best choices will be packages built around the open source AI splitter projects. It may already be that way.
Quality can vary dramatically depending on what source audio you're trying to split, the type of splitting you choose to do, and how well the model is trained. Nothing is magic or foolproof, and I think it's easy for some people seeing these for the first time to be overly impressed.
I'm sure it will be nice when high quality splitting is incorporated into all DAWs. I doubt whether it's a "game changer", but it's always nice to have extra functionality made easier to use.
Not sure why you think Spleeter is not "an AI one". It is an "AI one", same as all the others. E.g., read this old article about Spleeter:
This open-source AI tool quickly isolates the vocals in any song
There are certainly many people who pay for Moises, and think it's pretty good, and it's built on top of Spleeter. And notice what they choose as their web url: moises.ai
For splitter ai stuff, I discount any comments I hear like, "sounds like shit compared to [x]". You can hear other opinions where the same tool is extolled as "way better than all the others". Truth is it varies depending on what you're trying to do. From the looking around I've done, Spleeter and Demucs have improved a lot over last several years, and all of this stuff is improving rapidly. Current state of technology is that it's quite good at some things, not so good at others.
Just tested it... It is pretty good... And this is just the Beta... What is unique is, that it is incorporated into a DAW, and not just any DAW, but FLS... The things you can do without losing focus...
I use Serato Sample on desktop. It has the Stems splitter incorporated. I don't know if it's based on any existing code or if it's in-house from Serato. Anyone know?
It's pretty good. Generally the kick and snare come out clear, the cymbals less so, vocals and guitars mostly have artifacts. There's usually some bleed between the bass and drum tracks. Some files don't split well.
I'm not into sampling other people's stuff, I imagine it would be very useful for crate diggers and for sample based music. It's not mind blowing technology, but it's pretty interesting.
I haven't used it in a track, I don't know if I ever will, but I used it the other day to isolate a drum part I wanted to learn and it worked well for that.
Btw, there can be a difference in quality between apps using Spleeter. My understanding is that there are parameters that can be set that influence the quality.
Spleeter is an AI-based process. So, the distinction isn’t Spleeter vs AI but differences between different AI approaches or training sets.
Maybe flsm will follow?
It's all based around the same algorithm though right?
That's a significant oversimplification. We're not talking about a simple mathematical formula or piece of code that can be expected to have the same results wherever it's plugged in. Saying that is sort of like saying twelve students in three classes taught by three different teachers will all have the same test scores just because they all use the same textbook.
Many of the current crop of apps are based on the open-source Spleeter code. But that's just a code library. There can be many other variations under the hood in how they're used and optimized. And, not all are based on Spleeter. I believe some, such as Serato, were written before that was even available.
Maybe that depends on what you mean. Even if they were all using same research papers they could all have different implementations of base algorithms. I have seen the term 'spectral filtering' thrown around as applicable to most of them, though. Here's what ChatGPT (the free version last updated Sept 2021) said when I asked if it knew about algorithms used in track separation software:
==========================================================
Yes, the process of separating a stereo audio track into separate tracks for each instrument or sound source is known as audio source separation or audio source separation. It is a challenging task in the field of audio signal processing and machine learning, and there are several algorithms and techniques used for this purpose. Here are some common approaches:
Independent Component Analysis (ICA): ICA is a statistical signal processing technique used to separate mixed signals into their original source components. In audio source separation, ICA is applied to the mixed stereo audio signal, and it attempts to identify and separate the independent sources, which could represent different instruments or sound sources.
Non-Negative Matrix Factorization (NMF): NMF is a matrix factorization technique used in audio source separation. It assumes that the audio signal can be represented as a linear combination of basis vectors, where each basis vector corresponds to a source or instrument. By factorizing the mixed audio spectrogram, NMF can separate the sources.
Deep Learning Techniques: Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and deep learning architectures like deep autoencoders and Variational Autoencoders (VAEs) have been used for audio source separation tasks. These neural networks can learn complex patterns in audio data and perform separation tasks effectively.
Spectral Filtering: In this technique, the frequency content of the stereo audio signal is analyzed, and spectral filtering is applied to isolate specific frequency bands associated with different instruments. This approach is often used in combination with other methods to improve separation quality.
Phase Reconstruction: Separating sources based solely on magnitude spectrogram information can lead to phase inconsistencies. Some methods incorporate phase reconstruction techniques to improve audio quality and maintain the coherence of separated sources.
Semi-Supervised Approaches: In some cases, source separation can be enhanced by providing partial information or training data about the sources to be separated. Semi-supervised methods combine unsupervised separation techniques with additional information to improve separation accuracy.
Deep Clustering: Deep clustering is a deep learning technique that learns embeddings for different sources in a mixed audio signal and clusters them to separate sources. It has been shown to be effective for audio source separation tasks.
Score-Informed Separation: In cases where musical scores are available, algorithms can use the score information to guide the separation process, ensuring that each instrument follows the musical notation.
It's important to note that audio source separation remains an active area of research, and the effectiveness of different methods can vary depending on the complexity of the audio signal and the specific task at hand. In practice, a combination of these techniques may be used to achieve the best results, and the choice of method often depends on the characteristics of the audio material and the desired outcome.
I don't think it's accurate to call Spleeter a code library. Like Deezer [edit: I meant, 'like Demucs'], it compiles to an executable 'spleeter' command that's can be used in a command line interface to process audio files. I'm sure code in the project could be used as a library, but I suspect it's mostly used as a black box tool, with products using it wrapping the command line tool in a gui and probably modifying processing settings and training data.
Thanks for the clarification. My basic point was in the right direction but my use of terms was un-researched and inaccurate.
👍🏼
Thanks @hes : I wonder if the Stemz app uses Spleeter also .
They must be using some superior in-house developed alogithms that they built from the ground up. How else do they justify charging double the subscription price of Logic for Ipad? Sheesh! Nah, I'm good with the 'Stemz' that come with Koala thank you.
the cost is fractions of a cent to convert one song
Powerful cloud server=expensive. I would rather use my powerful M1. I can use it in Logic via Koala:)
Not all of these are built on spleeter, not all are good. from my intensive testing, lalalal's yielded the best results (cleanest / no artefacts). Serato's came in second as far as quality, but wins first place for ease of use / speed.
I suspect in another two years these will be so insanely good you will be able to extract every individual instrument with little to no artefacts
Interesting. I'm tempted to test lalalal and maybe purchase a 90 minute pack to see if it is that much better than Koala's or Transcribe's stem extractions.