MAISTERSTÜCK is a new AI enhanced system for algorithmic composition. MAISTERSTÜCK models all creative layers of a composer's mind and can create music from structural, stylistic, emotional and real-time attributes.
MAISTERSTÜCK is a service oriented API. You can apply for the here ↗closed beta
MAISTERSTÜCK can conserve, depict and recompose the characteristics (composition style, lyric style, sounds and playing techniques) of individual musicians (incl. singers) and their interplay in virtual bands.
March 3, 2023: Debut single
Guardians of Time of the world's first virtual AI metal band FROSTBITE ORCKINGS
in the style of Melodic Death Metal.
AI generated song structure, AI generated chord progressions, AI generated Lyrics.
AI generated arrangements for drums, bass guitars, guitars and keyboards. AI generated vocal arrangement.
Automatically created and pre-mixed audio stems of all instruments, including vocals.
(Mix, Master, Artwork and 3D visualization done by humans. For now, until the next update...)
All underlying sound sources and samples are copyright owned and exclusively produced by and for Musical Bits, the creators of MAISTERSTUECK.
MAISTERSTÜCK can create music in the desired style, with dedicated climax and turning points, for example as a video soundtrack.
Empowering Symphonic Soundtrack
TASK for MAISTERSTÜCK: For my 26 second video, create empowering music in the style of Symphonic Soundtrack with an intro from 0:00 to 0:09, tension building from 0:10 to 0:21, a pause at 0:22 and climax at 0:23.
Energetic Metal Soundtrack
TASK for MAISTERSTÜCK: For my 26 second video, create energetic music in the style of Metal with an intro from 0:00 to 0:09, tension building from 0:09 to 0:22 and climax at 0:23.
Melanchonic Piano Soundtrack
TASK for MAISTERSTÜCK: For my 26 second video, create melancholic music in the style of Piano Ballad with an intro from 0:00 to 0:09, tension building from 0:10 to 0:21, a pause at 0:22 and climax at 0:23.
MAISTERSTÜCK can create continous music based on a leitmotif (from user or AI generated) with dedicated sections of tempo and density, for example as a soundtrack for workouts.
For my 30 minute workout, create music based on my own Leitmotif with tempo sections 120, 140, 200 bpm.
MAISTERSTÜCK can create continous real-time music based on mappable real-time input, for example as your personal driving soundtrack.
For my cruise, create a real-time soundtrack with music in the style of Hard Rock based on Parameters Speed, Acceleration & RPM.
MAISTERSTÜCK can create songlyrics with given song and syllable structure & rhyme schemes and semantic colouration.
Highway to Hell
A verse and chorus fitting the song structure, syllable structure and rhyme scheme of AC/DC’s ‘Highway to Hell’.
Hymn of fire,
Fall of night,
Whither will she meet her in the night?
Still I see you,
Fate will wait,
We won't give up without a fight.
Down on my head,
We'll get you,
Wring my heart over an open sky.
The black death
Can break through
Because I don't believe the lie.
We try to bring back the dead
We try to bring back the dead
We try to bring back the dead
We try to bring back the dead
Love me Tender
Elvis’ ‘Love me Tender’ with a twist of heavy metal doom in the lyrics.
Don't let me stand in your way,
Then you will awake.
Long gone is the light of day,
But I'll never break.
And you're all I want for now,
Let him inside you.
No one has to show me how
we will make it through.
Free of death and free of war,
We begin the fight
where no man has gone before,
Away from the light.
You will know the end is near,
You are born to die.
Which way do you run from here?
Your life is a lie.
Can you make it on your own,
Man against machine?
Cause we were always alone
Can you hear me scream?
A little rock’n’roll cliché infused into the song structure of Black Sabbath’s ‘Paranoid’.
Lingering behind the hard night
Play it from the alleyway
It's your chance so baby try it
Eking out a way to play
For A Good Time
Make your move tonight
(Not so) Frequently Asked Questions
Is MAISTERSTÜCK Machine Learning? A neural network? The end of all musicians?
How does MAISTERSTÜCK work?MAISTERSTÜCK uses AI technology to model all layers of creativity of a human composer and implements these layers as reusable and combinable software components. MAISTERSTÜCK’s functionality can be accessed via a service oriented API.
Is MAISTERSTÜCK AI?MAISTERSTÜCK makes use of state-of-the-art AI technology, not only in its machine learning parts but wherever it serves the system’s purpose and its users’ demands. For instance, Markov Models are used as a base technology in composition and lyrics writing, and variational autoencoders (VAEs) are applied to increase variation in the system’s outputs on different musical layers, like drumming styles or the way musicians build their arrangements on top of each other’s input.
Is MAISTERSTÜCK Machine Learning?Our approach is to combine long-standing professional experience in songwriting and music production (as testified by multiple international chart-entries by the composers that work for and with us) with innovative domain-adapted machine learning techniques. Thus, some components of the creativity layers are handcrafted, while others work with patterns that were machine learned from data which we prepared for that task. Users of MAISTERSTÜCK are in control to decide how much they want automated or where they want to intervene in the creative process.
Is this Deep Learning from all music out there?In its machine learning parts, Maiserterstück learns specific aspects of a creativity layer from data specifically created by us, gathered and prepared for that purpose. For instance, style-typical chord progressions can be learned from available (collected or created) musical notation data into the chord progression layer. Thus, rather than using a “brute force” big data/deep learning approach from all available audio material out there, MAISTERSTÜCK makes an informed selection of data for its machine learning components.
What is a “creativity layer”?A creativity layer models a specific task that a composer or songwriter needs to tackle when writing, composing and creating music. For example, some creativity layers are concerned with the rhythmical structure (“the pulse”) of a song, other layers take care of the harmonic form, and yet another focuses on lyrics writing. There are layers for the interplay of musicians within a band context, others for adapting song structures to the expectations of specific audiences. Of course, different creativity layers are intertwined, such that the outcomes of one will feed into processes on others or feedback and re-iterate into earlier creativity layers.
Is MAISTERSTÜCK re-using existing audio material?Everything that MAISTERSTÜCK creates is based on audio sample material which we either completely created ourselves (together with producers who worked for international productions on gold and platinum record level) or have fully licensed.
Are you making composers unemployed?MAISTERSTÜCK can be used as a stand-alone-system (“compose music for me”), but it can also serve as a tool for composers seeking new forms of inspiration (“make me a suggestion for a variation of that chorus”). Composers, musicians, producers and computer scientists can also contribute components (sounds, models, algorithms) to MAISTERSTÜCK in different layers.
Who owns the copyright of the compositions?The output of the service oriented API is royalty free and belongs to the user that paid for it, as long as it is not re-sold. Complete productions, like the output of the virtual bands of the Metalverse, are copyrighted to the entity that composes the outputs of MAISTERSTÜCK to a full production.
So now computer nerds are in charge of writing music?The composers and producers contributing to the different layers of MAISTERSTÜCK participated in chart breaking productions worldwide, and worked for and with the biggest record companies, cinema production studios and game development creators. Their input is the base for any creative process that builds upon that.Thus, musicians remain in charge of shaping MAISTERSTÜCK together with software developers (who sometimes happen to be one and the same person).
Don’t you know that a computer system will never be able to fully replace a real great band / singer / drummer / songwriter / …?That is true (for now ;-)). And we think it is great. MAISTERSTÜCK can create great music that happily co-exists with all creative human output from the past, present and future.
Where are the limitations of the system?Most of the current limitations will disappear over time, as we are working on more and improved sounds, styles, models and creativity layers. The hardest part currently is to create outstanding vocals, but we think we are on a good path.
Can MAISTERSTÜCK create music / styles that do not exist yet?MAISTERSTÜCK is not designed to create scales, sounds, chords or comparable musical elements that nobody ever used before. But MAISTERSTÜCK can re-combine all existing elements in countless ways, and this can lead to styles and sound experiences that do not exist yet.
When can I use the MAISTERSTÜCK solutions?You can apply for the closed beta here ↗
What are you working on next?We are going the full path to virtualizing bands and musicians, including their characters, their social media appearance, their live-on-stage acting and much more.
I find this cool and want to engage - how?We are always on the lookout for dynamic individuals and organizations to join forces with.
If you are a professional in AI and/or music, please take a look at the Job descriptions on the Musical Bits website.
Apart from official jobs, whether you are a visionary in your field, seeking to enhance your business, or simply looking for the right partnership to take your endeavors to the next level, we would love to hear from you. We are eager to hear from you, so don’t be a stranger and reach out to us via: info@Maisterstueck.com
Who and How
MAISTERSTÜCK is a product of Musical Bits and was created by a team of wizards with megalomaniacal ideas.
Armin, Bastian, Charlie, Chris, Denise, Dennis, Georg, Gerrit, Hagen, Inga, Jens, Jörg, Kurt, Marten, Nina, Robin, Stefan, Thomas, Tomasz.
Bayes Classifier, Cellular Automata, Decision Trees, Generative Adversarial Networks, Generative Grammars, Inductive Statistics, Latent Semantic Analysis, Markov Models, Rule Based Hierarchies, State Spaces, Support Vector Machines, Tf-Idf Scores, Transition Networks, Variational Autoencoders (and a lot of coffee, love, indisputably bad jokes plus tons of analogue and digital studio gear).