What is PhonicMind's Stems maker?

PhonicMind was started as vocal remover service in 2017. Two years later we evolved in to Stems maker. All methods used up until 2017 trying to remove vocals where based on phase differences of stereo track and some frequencies custom selection. Some academics even tried to figure out repeating patterns in the song and separate vocals that way, as vocals usually do not have repeating pattern. None of these methods really worked well enough.

PhonicMind's vocal remover opened a door to the new era of sources separation by introducing AI based vocal remover, which outperformed every other method used before.

In 2019 we have evolved in to Stems maker, which not only separates and removes vocals form any given audio track. It also removes drums, bass and other instruments in to separate stems. Again by using latest and greatest AI techniques.

Basically our artificial intelligence understands music. We are not sure if it can feel it, but it clearly understands it and knows what to eliminate and what to leave.

We are also continuously pushing it's capabilities up to the new levels on daily basis. So it's getting better and better over time.

How to make karaoke online?

From now on you no longer need to wonder on how to extract vocals from a song. You no longer need to fiddle around with Audacity or other audio editing tools trying to reach impossible. Separating vocals from full song was always a tricky business even by using greatest audio editing tools. It's a tricky business not only for daily users but even for audio professionals.

Simply upload your .mp3 file and we do the conversion online within the minutes by using our state of the art sophisticated artificial intelligence. Once the process is done you will be able to download karaoke (instrumental) track for your fun or project.

Whatever it is - pop, folk, house, electro, rap, rnb, - you name it. Whatever language it is in. It all works equally well.

What can I do with PhonicMind's Stems maker files

It's totally up to you what you do with those files. Most of the times people just want to make a karaoke version of their favorite song so that they could sing or play the main melody of a song with full music background. A lot of the times people do these conversions to sing a song in a different language. That is very fun!

Eliminated vocals, drums, bass or other instruments are usually used by music professionals to create outstanding remixes, by re-sampling some older song and giving it a new life. That is very beautiful! If you are one of those guys, please share your work with us, we would love to see what you are able to achieve with this!

You can also download .stem.mp4 file of your song. This is file format created by Native Instruments. By using .stem.mp4 files with Native Instruments Stems products you can do DJ'ing like never before!

If you are using or willing to use it in some other way, please share your thoughts with us, so that we could maybe make our vocal remover better suited for your needs.

What is a .stem.mp4 file?

Stems (.stem.mp4) file is an audio file that contains a track split into four musical elements: A drums stem, a bassline stem, a harmony stem, and a vocals stem for example. The Stem file also includes the original stereo master of the track for standard playback.

Experience new era of DJ'ing by using our Stem files with Native Instruments:

What are the requirements on file that I have to upload?

Our Stems maker accepts all major audio file formats like MP3, AAC, WMA, FLAC, ALAC and many others. The better quality audio file you have, the better end result will be. We recommend using CD quality audio files for best results. Your song file should be less than 100Mb in size and shorter than 9 minutes.
Attention! You might not get the best results out of our vocal remover if you use poor quality audio files.

What is your output file format?

Our AI Stems maker engine works with CD quality audio in it's brain, therefore we always give you CD quality (44.1 kHz, 16-bit) high resolution files back as end result.
If your original file is in lossless file format like .wav, .aiff, .flac or .alac, we will give you back lossless .flac files in return as well.
If you upload your song in any compressed format like .mp3, .wma or other, you will get high-res 320kbps .mp3 files back.

Do you have some examples?

Yes we do have examples - please find them on our How it works page.

Is there a downloadable PC application that I can purchase?

No, there is no downlodable application.
All of this work is being done with deep neural network which is running on an expensive GPU cluster that is organized in a very specific way.
In other words - regular PC does not have enough resource capacity to perform this, that's why there is no downloadable PC application and we do not see that it could become available in a near future.

How about some subscription based pricing model?

We are very glad that we are constantly getting this question, but it is not possible at least at this moment in time.
Every conversion has very high computational costs associated with it, therefore if we would allow people to do unlimited number of conversions over some subscription period, it would just over-flood our GPU cluster or even burn it out. We are unable to offer this at least now.
Although we think it's likely that we might offer this one day, as our AI evolves enough and GPU servers get's cheaper over time.

How many free preview files is there available per user?

We allow 20 preview files per user/session and we add 20 preview files per each credit purchased. So for example if you exhausted all of your free previews and purchase "Basic bundle", which includes 5 conversions, that will add 100 free previews on to your account as well.

Is there money back guarantee?

Sorry, but no. Every conversion has high computational costs associated with the job. Please review our examples and perform tests before you make any purchase to get familiar with what kind of result you might expect. Usually conversion results are the same or even better quality than provided examples.