Free Plugin, Samplab, Lets You Edit The Notes In Polyphonic Audio Samples

Swiss-based music software company Samplab has introduced Samplab, a plugin that lets you edit and rearrange pitches in polyphonic audio samples.

Here’s what the developers have to say about the plugin:

“When you drag and drop an audio file into the plugin, it is sent to our server for processing. Once this is completed, you recieve the result with all the notes in the audio separated. In order to edit the notes, we generate a MIDI file that you can drag into the Samplab track.

From now on, the plugin acts like a virtual instrument: It gets the MIDI events as input from your DAW and plays the (edited) notes. If you don’t make any changes to the MIDI, the result should sound the same as the original audio. Of course, you can also just use the MIDI file for other things.”

Pricing and Availability

Samplab is available now as a free download, for Mac & Windows.

Note: Samplab is currently limited to editing samples with a maximum of 127 notes.  Because this is a client-server app, an Internet connection is required.


33 thoughts on “Free Plugin, Samplab, Lets You Edit The Notes In Polyphonic Audio Samples

    1. Wow? You can hear a lot of artifact sand indeed this short example has a lower quality than Melodyne or Zynaptiq.

      The Cloud Is Just Someone Else’s Computer.

      if you’re not paying for the product, you are the product.

      I hope the two friends making this thing will get hired soon.

      1. Yes, Ludwig, WOW.

        Yes, it has lower quality than a product costing $850.

        Yes, we know what the cloud is.

        What’s wrong with being a product? I bet you own clothing with company logos on it and don’t think twice about that. And you PAID to be their product.

        I hope the two friends making this thing are having more fun than you seem to be.

      2. That saying is pretty inadequate. In most cases of surveillance proxy products you’re akin to a resource node, like a tree, or an iron node..

        This doesn’t appear to apply to this product like it would to, say, your Facebook or Twitter usage. Here audio is processed through cloud services running on Google cloud temporarily. The rest of the data seems they collect seems to be standard security and cookie-esque data (someone correct me if I’m wrong I’m just looking at the privacy and terms of use tabs)

  1. This is a great idea but the plugin should access a local server installed on the host computer IMHO. There is already enough “call homes” from plugins for my personal comfort level and also prefer to work on music completely disconnected from the web unless I’ve dug myself a hole and need access to tutorials or to update software, but otherwise, too many distractions for me. I wonder if this is how they are going to train their pattern recognition system before releasing a paid version, and if so, how much the full price version would be? Regardless of al that, the sentiment behind it is great. This plugin definitely has my interest.

  2. What a fascinating and cool concept. It’s like using Melodyne’s DNA (direct note access), but using the extended power of cloud CPU to do a non-realtime process.

    This has some interesting im/ap-plications.

  3. Question for the techies – why would this process need to be cloud-based, when an application like Melodyne does its pitch-editing on your computer?

    Does this somehow require much more memory or processing power?

    1. Without knowing the processing costs of samplab it’s hard to say with total confidence, but I imagine cloud computing makes quicker work of materially and musically useful polyphony isolation to editable midi back to the sample manipulation.

      The demonstration was pretty quick and musical with little artifacts, some FFT DSP wizardy. That would most likely tax normal CPUs running a lot of other DSP. In this way you can have their server do all of the heavy lifting of the program for you so it’s not costing a lot of space in terms of other DSP in your project.

      And if it is that cpu intensive, then cloud computing also makes samplab accessible to more people by lowering the use requirements to internet connection rather than a powerful CPU.

      1. Thanks for the feedback.

        To me the server dependency of this app is a big downside. I know at least that iOS devices and Macs have built-in processors specifically for machine learning tasks, so it’s not really clear to me what makes a job require a cloud-based server vs your device’s capabilities.

        I agree, thought, that the technology demo is pretty impressive. I’d like to see more audio examples or see how people use this. Seems like it could be a great thing for remixers or anybody that uses samples.

        1. In short, your computer could maybe task this in a DAWless environment, but in parallel with reverbs, delays, instruments, etc this function in a plugin may not be viable to many people without the cloud based computing. The cloud computing ensures the parallel processing function of this.

          I think about the make up of a single guitar string plucked and there’s the resonant frequency of the fundamental, the transient of the note being struck, the overtones, and the energy decay of all of those things, for a single note.

          Where that most likely gets expensive CPU wise is abstracting multiple transients, fundamentals, overtones, etc and in a musical way and, while a Mac or iOS device might have processors specifically built for machine learning tasks I think the novel use case for cloud computing in this environment is that it allows this algorithmic processing to be done in parallel with the rest of your music process untaxed.

          It’s very impressive! Yeah I’d love to test the use cases people have been questioning in the thread, I’m just in a class at the moment so I can’t, but it’s free so maybe you could do it!

    2. I would guess that they use machine learning based algorithms. These often need huge data corpora and/or GPU processing power. It wouldn’t be very practical to download & install them on typical PCs used as audio workstation.

  4. A good concept.
    How accurate is it with polyphonic , multi instrument music pieces!

    I prefer to have the analyser / engine on my PC, so to not need to use an internet connection + data.

  5. Clever and potentially useful.
    Sure, much of it sounds weird like overstretching samples might. Still a nice way to transcribe polyphonic snippets and slice samples “vertically”.
    Doesn’t work so well in Bitwig Studio 4 on my Mac (unable to drag audio files in or MIDI files out).

    1. My audio computer is a Mac and always connected to the internet. Never had a virus or a stability problem either. Plus I don’t have to switch to a second computer to be smug about it on Synthtopia 🙂

    1. A lot of people use polyphonic audio samples. Not everyone is great at sound design and some people want to create sound design using premade sounds.

      There’s a ton you can do to make a sample your own in post effect processing, so this technology has merits on that ground.

      Beyond that, it can be used informatively. I play music with a guitarist who doesn’t know formal theory. He can send me stems that I run through this algo to gain quicker insight into his song structures via midi without him fumbling through explanations.

      I can rearrange his chords and show where we can make subtle changes in his recorded chord progressions to inject more interest into his phrases with his own timbre fairly effortlessly.

      There are probably other use cases I’m not thinking of but there’s differently a good amount of potential.

      re: data, It looks like I’m getting a useful tool for my data ingested into their algorithm to improve their product, probably for refinement to create a sellable product in the future. The trade off feels pretty fair here.

      The devil is in the details of the terms of service or end user license agreement, but on face value the value proposition feels pretty even.

  6. wow need to try this, could be fun to play with and see what it does to random non melodic samples – I like doing that with the midi detection in ableton

  7. Really quite interesting. Currently limited to a ten second sample. Sample drag and drop doesn’t work in Logic Pro/Catalina but you can manually load the sample through the plugin interface, and the midi drag and drop does work. Stores the sliced samples in ~/Library/Samplab_Data/(date & timestamp)/. I guess you have to save your project because if you swap out the plugin to try the generated midi file in another instrument the sample mapping for that track is lost. Definitely some artifacting when you examine individual slices but less noticeable when played together and close to the original pitches. I’m impressed and think this could be useful. Look forward to seeing how the offering evolves.

Leave a Reply

Your email address will not be published. Required fields are marked *