wizardofid
Honorary Master
- Joined
- Jul 25, 2007
- Messages
- 12,540
- Reaction score
- 6,812
It is what I sort of been doing but the manual hard way of manually setting it up. That said I still prefer my way, it keeps all the previous setup and with every new chat I setup some new standard rules only grunge only 4/4 and for this chat only. Since I experiment a lot I am not exactly sure how rigid your system is. But my setup basically give me the lyrical tag descriptors the tagged lyrics and then the description.
Had to manually setup it up through trial and error, i have also setup global rules unless I ask for specific effects it doesn't add them just gives me the default output. For the afrikaans I needed to setup the rules specifically for that chat language rules ect, ect.
You system works great I am not so sure it works well for tinker in me. Like setting up custom chats they way I get exact output I want, it follows then all the rules in that specific chat, really useful, if you want to use a specific style, voice or instrument. Actually discovered a couple of new lyric tags reverb, and then bloom, tail, delay, stutter. So It would be [reverb bloom] [pitch +2] for a specific line the while verse or chorus or word specific.
My system now it very much plug and play now minimal effort. Now I just have to describe the lyrics, what it is about give specific lyric direction and simply tell it give it dynamic range between verses and chorus make it female or male and it tags it without and issues. I have also added chords, previously I have added the chords in the description or top tags but recently changed to chords per line.
Yeah but it brings me to the next point of the assumption of all people use the same lazy approach to generating music, and sure there are folks that simply plop down some stuff and job done, it gives the folks that actually sit down and spend hours tweaking their input a bad name really.
Suno have really stepped up their game since model v4.5 it is much more of DAW experience now than previously, without all the nasty wires hanging out.
V5.5 already blew my mind with the micromanagement of the output already and the serious improvement to vocal processing. model 6 is going to be absolutely next level. You can already sort of do the same vocal processing to instruments that you can do with voices it is still sort of hit and miss. But I would love to see more processing of instruments. Something I found with grunge, metal or and electric guitar effect the range is pretty basic, and guitar feedback is hit and miss and when it does it is very mechanical it does a bit of overdrive, but still rigid.
Haven't really experimented too much on that front, but honestly it would be a nice touch to have better control of that. The other thing I would like to see guitar licks it is very generic at best, suno mostly avoids them. But something I need to experiment with some more as well.
