Site Overlay

In 1958 Pete Seeger recorded Black is the Color. I revisited this folk song with the help of an artificial intelligence developed with researchers at Spotify CTRL Paris. You can listen to the original recording from the catalog of Folkways Records.

The tool created an original harmonization on Pete Seeger’s voice

How it works?

The AI tool works with a target and a source. The target is the melody that you want to harmonise and the source is the piece of music that the tool will adapt to the target.

For example, I used a string score from a Bossa nova song (very different from Black is the Color). I wanted to feed the machine with very sophisticated harmonies on the simple melody of Black is the Color.

SKYGGE used Bossa nova to flesh out a folk song…

The AI technology adapted the Bossa Nova strings score to the melody and generated a great string score with beautiful harmonies, like if it was created for this specific song. A string quatuor played this arrangement in a studio.

… and Elizabethan era music too

The otherworldly juxtapositions include Canadian musician Kyrie Kristmanson’s Amazing Grace enriched with the adapted sound of William Byrd’s mass for three voices. I used a piece of harp music from the romantic period turned into an acoustic guitar on Pete Seeger’s Shenandoah.

Elsewhere, artificial intelligence generates a choir from Pete Seeger’s voice on Black Is The Color, multiplying it into a surging wash of vocal harmonies. An AI-generated sax solo provides a locked-in groove for Amazing Grace; and an AI-generated trumpet solo carries Peggy Seeger’s Song Of Myself into experimental electronica territory.

Inspired by these new AI-generated ideas, I recorded his new EP with the French double bass player Stéphane Kerecki, and a string quartet performing these AI arrangements.

Here, several videos in which I describe this unique experience of creation with the prototype tools for harmonisation and smart sampling.

7 thoughts on “American Folk Songs : A.I. and harmonisation

  1. We believe we have achieved this goal. All the songs in this album have a special character that makes them uniquely singular! Additionally, for unknown reasons (perhaps due to an optimal exchange between stylistic consistency and innovation produced during the AI generations) most of these songs are is the first album composed by an artist SKYGGE with artificial intelligence. Its goal is to show that AI can be used to create new, compelling music, generating fresh musical material of all sorts: melodies, harmonies, timbre, rhythms and the like.

  2. Hello,

    I am impressed by the results and was wishing to use AI for making music, with an interest in Flow Machines.
    I would love to spend some time with you to see how you use, if possible, this technology and how I could do use it as well.
    Congratulation for the Hello World Demos EP. For this one, would you share how the voices were created? Do you use a Text to Speech solution?
    Thanks a lot if you take some time to answer my message

    1. Thanks Patrick for your message. Flow Machines owns to Sony CSL, you should reach them if you want more infos! Voices of Hello World Demo are created the same way that the album, but there is added sounds, productions on it.

  3. Hi,
    I’m producing a radio show and would like to ask you just a few questions about the “technical” way you make your music, as I plan to broadcast one track from your CD “American Folk Songs”. Could you reply to my email address maybe ? Many thanks. Best regards.

  4. Chaeyeon KIM says:

    Hi, there!
    My name is Chaeyeon KIM, who is planning graduation peice(Motion Graphics) in this year☺️
    Can I use your song ‘Cold Song-Composed with AI’ for my graduation peice? Cause my title of graduation peice is ‘AI’.🤖
    Of course it’s not commercial thing, I’m going to use it for my graduation peice and later just on my personal portfolio.

    Your every work is MIRACULOUS !!

    I look forward to your reply 🙂

Leave a Reply

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