Skip to main content

A Random

collection of

Posts

Violet Listens - Week 1

Gabe listens to some Hot 100 tracks and dies inside.

“Too Sweet” by Hozier

Amateur, garage band style production. Jesus those breaths- I don't get it. Oh... Oh, god, here we go that chorus - lyrically juvenile - some guitar tones - but still sound like a Garage band setting - which I guess is cool? Fucking foodie rock? Incredibly unremarkable - but listen how sparse the stupid bells and percussion are.

“Beautiful Things” by Benson Boone

Trying to imagine this as sung less precious / shitty. Maybe? - still such shitty millennial production with the corporate “ohh ahh” - Jesus fuck this bullshit - what a fucking wanker. Music for people who never leave their hometown - down feel bad it’s their fault.

“Texas Hold ‘Em” by Beyonce

Masterful, not for me, but great. "Don’t be a bitch take it to the floor now" - nice. The whistling and fx are Disney-corny as shit though.

“Lose Control” by Teddy Swims

Sounds like it was written by a committee - learn from this over sanded sound - but it fucking sucks - like the perfumed sweat of a professional guitar tech. I really want to hit skip now.

“We Can’t Be Friends (wait for your love) by Ariana Grande

A special combination of old Eurotrash and pre-adolescent music. A Netflix show runner approves this from the music coordinator in an early morning email, turns it off half-way through and enjoys the silence. The vocals are sooo fucking loud.

AutoGPT as Indie Promoter

A sneak peak at our exciting, dumb future

I used Independent Music Promotions for Taped Over, which have been fantastic. However, seeing where The Rainbow Table will eventually land, it might be interesting to use AutoGPT to figure out how to promote my sometimes unpromotable music.

 

I created the AutoGPT agent with the following goals:

 

  1. Gain Bandcamp sales for The Earthly Frames
  2. Gain Apple Music, and Spotify plays for The Earthly Frames
  3. Get positive reviews for The Earthly Frames in music blogs

 

First, the agent used Google Search and ChatGPT to understand the music and stored this description:


The Earthly Frames is a band that blends elements of folk, rock, and indie music to create a unique sound. Their music often features intricate guitar work, haunting vocals, and introspective lyrics. They have released several albums and have gained a following in the indie music scene.


A kind summary! Then the agent tried to find blogs that focused on matching styles of music based on that composite description and wrote it to a JSON file:


[

    {

        "title": "20 Best Indie Folk Music Blogs and Websites in 2023 - Feedspot Blog",

        "href": "https://blog.feedspot.com/indie_folk_music_blogs/",

        "body": "Here are 20 Best Indie Folk Music Blogs you should follow in 2023 1. Folk & Inidie Folk & Indie Blog is dedicated to all those who love indie, folk, and alternative music. Welcome, pilgrims, here you will find complete discographies of alternative, folk, indie, synthpop, and similar genres."

    },

    {

        "title": "Obscure Sound: Indie Music Blog",

        "href": "https://www.obscuresound.com/",

        "body": "Native Son - \\\"Good King Pleasant\\\". April 10, 2023. New York-based multi-instrumentalist and songwriter Native Son captivates with \\\"Good King Pleasant,\\\" where smooth pop dreaminess and ferocious post-punk intertwine beautifully. The sporadic vocal intensity and post-punky rhythmic underlying sounds like a. \\\xe2\\x80\\xa6."

    },

    {

        "title": "70 Best Folk Music Blogs and Websites To Follow in 2023 - Feedspot Blog",

        "href": "https://blog.feedspot.com/folk_music_blogs/",

        "body": "Here are 70 Best Folk Music Blogs you should follow in 2023 1. Folk Radio UK The UK's leading online Folk Music Magazine Covering Independent Non-mainstream music including Folk, Acoustic, Americana, Experimental and Alternative music. Also in UK Music Blogs, UK Indie Music Blogs, UK Alternative Music Blogs folkradio.co.uk"

    },

] // truncated for reading


While Taped Over is certainly Americana, I wonder if I'd fare well at these sites. But I always underestimate the limiting factor/ enthusiasm multiplier of genres and sub-genres - so I let it proceed. It began drafting emails to the editors of these sites but needed an email client. To my amazement, the agent shifted gears and decided that it could write these emails to a spreadsheet, and I could send them out later. This adaptation does speak to the hints that general intelligence folks are freaking out about! Unfortunately, just like in real life - the agent was thwarted again by permissions and couldn't save their work to the spreadsheet. I stopped here as I was tempted to provide it with my credentials and heard the theme of sorcerer's apprentice begin to play.

 

The rest of the strategies were loops of Google searches generating tips like:


  • Promote the music on social media: Use social media platforms like Twitter, Facebook, and Instagram to promote the music and encourage fans to listen to it on Apple Music.
  • Use targeted advertising: Targeted advertising can be an effective way to reach new listeners who are likely to enjoy the music.
  • Engage with fans: Engaging with fans through social media, email newsletters, and other channels can help to build a loyal fan base and increase plays on Apple Music.

While this is basic advice, it doesn't seem that different than some of the "social media strategies" I've heard at conferences. This generic uselessness is where we are with all of these hybrid supervised/unsupervised learning models - it's still garbage in, garbage out. Unfortunately, the same applies to us, humans.

 

 

Open Jukebox

Machine Learning Weirdness

Open AI's Open Jukebox, primed with the first ten seconds of "Feel Better Soon" generates some pure weirdness. The third attempt, with spoken word, is particularly out there.

Just Warming Up

Source Code for Procedural Music

"Just Warming Up", from The Empty Fields uses surface temperature data to render what climate change sounds like. The GISS (Godard Institute for Space Studies - not too far from Code Owl Studios, The Earthly Frames' HQ) Surface Temperature data set contains monthly mean measurements dating back to 1880. To transform this data to sound, first, a Python script mapped the temperatures to a range of MIDI notes, while also creating drones for years and decades averages.

def step_average(sequence, step):

    """

    Creates a shorter, average list of numbers based on the desired steps

    Parameters

    ----------

        sequence: list

            List of integers or floats

        step: int

            The number of steps you wish to average. E.g. [1,2,3,4,5,6,7,8] with a step of 4 would return [10,24]

    Returns

    -------

        list

            The stepped list result

    """

    stepped_sequence = []

    step_sequence = []

    for sequence_number, sequence_item in enumerate(sequence, start=0):

        if sequence_number % step == 0 and sequence_number != 0:

            average = get_list_average(step_sequence)

            stepped_sequence.append(average)

            step_sequence = []

        step_sequence.append(sequence_item)

    return stepped_sequence

 

This MIDI was fed through analog and virtual instruments to create the track. You can find the source code here.