Description of the process by Daniel Sommerfeld.
Below there are some links to songs, which I have created combining two AI apps. These songs are my own property and may be shared if my name and my email address are clearly visible together with the song.
Daniel Sommerfeld, dani.sommerfeld@t-online.de
Some blind people seem to find it complicated to create music, using AI models to assist them. It indeed looks complicated but once you understand what a generative AI is and how it does what it does, it is so easy.
First of all, let me try to explain what a generative AI like ChatGPT or Suno is.
A generative AI works with your input. The more, and detailed, input you give it, the more it gives you back as output. That can mean that you have to type a novel like text in one message, but that is totally ok. The more you write, the more an AI gives you. That is, why those models are called generative. If you think that an AI has its own creativity, try to take distance from that idea. An AI has a basic knowledge, which it uses. But, if you want detailed information about something, you always have to tell the AI to search all available sources on the internet. To make sure you have the latest info, just ask it something like, get me the latest information, or, how old are these results?
It can be fun to communicate with an AI in the style of a dialogue and if you don't need exact and detailed info, it really can be fun but if you have something you want to work on using ChatGPT, it is always good to write long messages and to try to put as much details about what you want in it.
I was interested in creating songs using Suno. Once I got used to the website using NVDA, which indeed works better with NVDA than with JAWS. I started asking myself, how to keep all the controls in mind that Suno needs to create a song for me.
But talking about this with ChatGPT showed me that it was so easy.
Let me talk about the parentheses first and then tell you more about how to work on creating the right lyrics and style tags using ChatGPT.
Basically, you only need two types of symbols, parentheses and brackets.
The first one is being used for intonations and the second one for parts of the song like:
[Verse 1]
[Chorus]
[Bridge] etc.
If you want special intonations, it would look like this:
[Verse 1]
(Sweet)
[Chorus]
(dramatic)
[Bridge]
(Flute solo)
Yes, you can use that one also for instruments, you want to hear at certain parts.
It of course works best if you have an account, so ChatGPT can keep things. This also works with the free version but of course not as good.
But for Suno things, it should be ok.
So the first thing you should do once you found your way of working with the parentheses is to set a rule for ChatGPT like:
Always when you help me with Suno and give me something be aware of this. Choruses, bridges, breaks and so on are always in between left and right brackets [ ] Intonations, spoken etc., always in between left and right parentheses ( ).
Once you set it up like this, there is a good chance that ChatGPT keeps this template and follows it in the future.
And now you can do something, which I find very interesting.
Instead of searching Suno's help section, you can also go through it with ChatGPT because the help section of Suno does explain things but not always very clearly. Asking ChatGPT about certain parts and telling it to search the net about that will give you a lot of information, which you can either copy or you can also tell ChatGPT to create a text file for you, which you can download and use it as a reference.
And you can also use ChatGPT to create lyrics using your input.
Not like:
"Give me lyrics for a country song about two huge trucks."
You can describe a situation and then ask it to create lyrics out of this info you handed to it.
And then, in a follow up, you can always fine tune the lyrics until you
say: "Yes, this is it."
And a little hint:
You can tell ChatGPT to mind words, AIs like to use the same words a lot like "vibes" etc. That can be bothersome. *Grins*
OK, my fingers start hurting. I am not a good writer but I hope, that gave you a bit more understanding, of what one, blind or not, can do with AIs.
And now, as promised, a few songs for you I made using Suno AI.
Each song has a play button, mute button, a slider control and a button labeled: "show more media controls more options". If you activate the show more media controls more options, download is one of the options.
These songs are my own property and may be shared if my name and my email address are clearly visible together with the song.
Daniel Sommerfeld, dani.sommerfeld@t-online.de
I told AI Music Gen to use a guitar, Balalaika and accordion to give the song a Russian sound:
A version done with Suno AI, but it ignored my instructions about guitar, balalaika and accordion completely and followed its own "head". It created a nice song, but totally lost the Russian feel.
This is another Yevtoshenko poem, but I translated it into Afrikaans. I then requested Gemini AI Studio to generate me an instrumental track with a Russian sound using a balalaika and an accordion which it did and created a piece of music of 30 seconds long which it called Tears of the Taiga. I then gave that piece of music to Suno and asked it to use it as the basis to build my song on.
IN English, the poem is titled Colors. The English version is below the song.
When your face came rising above my crumpled life,
the only thing I understood at first was how meager were all my possessions.
But your face cast a peculiar glow on forests, seas, and rivers,
initiating into the colors of the world uninitiated me.
I'm so afraid, I'm so afraid,
the unexpected dawn might end, ending the discoveries,
tears, and raptures, but I refuse to fight this fear.
This fear-I understand- is love itself.
I cherish this fear, not knowing how to cherish, I, careless guardian of my love.
This fear has ringed me tightly.
These moments are so brief, I know,
and, for me, the colors will disappear when once your face has set
... Translated by George Reavey
A poem by the ancient Greek poet, Sappho, set to music by AI Music gen. Christo instructed it to use a bouzouki and a female singer who sings in a Greek style to give the song an old Greek sound.
a poem of which ChatGPT wrote the text and set to music by Suno. Christo instructed ChatGPT to generate a villanelle which is a particularly complex form of poetry with a fixed rhyme scheme and repeated lines. ChatGPT figured that out perfectly.
Created on June 6, 2026