Maybe it is something that may work for you, maybe not at all, but I can share my own path.
I am a computer engineer with no music background other than the basics I learned in music classes in high school 30 years ago. I do listen to music a lot though.
I have been working for a while now writing a story, and a few months ago I had the idea to try making a theme song for my story. What initially started out as a single song ended up being an entire project with the lyrics for 22 songs written so far.
Since I am no composer and I don’t have the knowledge or experience for that part, I used Suno (generative AI) to bring my lyrics to life. With persistance, sometimes generating dozens upon dozens of takes for each song until I got THE one that I just want to listen in a loop all day long, I got pretty good results.
Now, while the Suno songs do sound very good in their own right, I am considering them as prototypes for the final songs. There are technical reasons (some glitches here and there, voice consistency across the songs and some tweaks to the vocals) as well as wanting a final track that, while composed by AI, is not in itself AI-generated.
My next step is then to split the stems (separate vocals from instrument), use the vocals stems in Synth V to convert to a voice track using the voice-to-MIDi feature. This gives me a good starting point for the synth V vocal track, but still requies work to sound right. Once I am satisfied with the vocals track, I mix it back with the instrumental stem in my DAW. For that step, I use some plugins on the vocal track for voice shaping / post processing. A user here metionned a plugin called “Vocal Menace” which combines compressor, limiter, equalizer, reverb, delay and drive all in one easy to use plugin, and this is what I use now, combined with a voice multiplier and a final equalizer.
This is the step I am at now, learning on the fly how to use Synth V and a DAW.
Then, my plan is to re-build the instrument tracks using virtual instruments in my DAW, so that the end result is a human-made cover of my suno-generated prototype. I will have to learn on the fly how to do that. I found one software that would be a huge help in this process: RipX. It can analyse the song and decompose it into layers for voice and several instruments (piano, guitar, bass, strings, percussions, etc) and places all the individual notes on a piano roll. From there, you can see the tempo, the keys, guitar chords (for guitar tracks it gives you a neat drawing on top showing the guitar chords with wich fret to use for each string). It can also export these layers to MIDI. Great tool for transcribing the Suno-generated song.
Among the notions I’ve had to learn so far, there was the basics of using a DAW, of using Synthesizer V, but also how to post-process the Synthesizer V vocal track in my DAW (voice shaping, mixing) so you might want to look into that.
Just as a real singer’s recording, the vocal track generated by Synth V might not sound at its best in your song without proper voice shaping and post-processing, this can realy enhance how your voice track sounds.
As for how to learn these things, I am a self learning quick study, so I jump right in, and whenever I hit something that I don’t know how to do, I look it up. For this, Google, YouTube and ChatGPT are good friends.
This is my path, and for me using Suno is a good choice, because otherwise my project would never have seen the light. For me it opened a door that was previously locked. It might be a good choice for you, or not at all, depending on your willingness and opennes to use a tool like Suno vs composing your songs yourself.