Grok's Notes: Uploading Vocal Tracks (May 2025)

+ Show full styles

V4.5

Hey, it's been a while.

I've still been active on Suno - just mostly doing private, unpublishable songs. I'm writing a musical with Suno, you see. It's not gonna be shown here - it can't, all of it has my voice on it as the base track (pre-remix). And since it's a fan musical and involves copyrighted stuff it won't be available elsewhere, either. But people are listening to it elsewhere and loving it, so that's good enough for me. Well, I guess if it's done I'll put a link up somewhere so folks here can listen to it too, but for the moment, nah.

So, some interesting things I've found while uploading songs to Suno.

1. The more musical your voice is, the less autotune Suno will use and thus the less it'll stray from your original intention. That being said, it's still something of a dice roll - if you stray away from the song structure that Suno likes, or the genres you've chosen, it's still going to go a bit funny.

2. If you DO however stick to the structure, even past the limit of where your track has (this was done before the 8 min increase in uploaded audio), you can actually still get a pretty good roll of stuff that sticks roughly to what you've done before. Obviously, the more rigid your structure, the more likely it is for Suno to repeat it. It's still gonna take credits, just less.

3. If you upload a track that has a lot of ambient noise in it be prepared for Suno to sizzle you with static. There's too many points for the AI to pick up, so you're just going to get a really messy output.

4. (This is a sidenote to uploading vocal tracks - the usual route of making electroswing manually is now POSSIBLE but takes like four covers minimum. Electroswing manually is - take a sample, usually jazz, then cover it with a house beat. I did this with Workman Swing, so it's now POSSIBLE. Just.... needs a lot of adjustment.)

====

5. v4.5 is an interesting one, compared to the previous models, imo. It takes full sentence prompts now, and they can be EXTREMELY specific. I see a lot of experimentation in the future - long ago, when I first started, I used to play with using marketing copy in the styles box. This might work again, but INTERESTINGLY this time.

Well, SSC7 is coming up soon, so I think I'll be sticking to my usual guns for it this time unlike last year's ambient liquid dnb/rock - electroswing + melancholy poetry (this is a preview of the vibe...except less techno). I should also maybe try out an alternating duet since I know how to do that cheaply now (see the previous notes for that). And maybe do a rap/sung combo? Geez. I've experimented a lot the last couple of months, that's cool I think. Well, see you guys next month (hopefully).

2695/7500

Hey, it's been a while. I've still been active on Suno - just mostly doing private, unpublishable songs. I'm writing a musical with Suno, you see. It's not gonna be shown here - it can't, all of it has my voice on it as the base track (pre-remix). And since it's a fan musical and involves copyrighted stuff it won't be available elsewhere, either. But people are listening to it elsewhere and loving it, so that's good enough for me. Well, I guess if it's done I'll put a link up somewhere so folks here can listen to it too, but for the moment, nah. So, some interesting things I've found while uploading songs to Suno. 1. The more musical your voice is, the less autotune Suno will use and thus the less it'll stray from your original intention. That being said, it's still something of a dice roll - if you stray away from the song structure that Suno likes, or the genres you've chosen, it's still going to go a bit funny. 2. If you DO however stick to the structure, even past the limit of where your track has (this was done before the 8 min increase in uploaded audio), you can actually still get a pretty good roll of stuff that sticks roughly to what you've done before. Obviously, the more rigid your structure, the more likely it is for Suno to repeat it. It's still gonna take credits, just less. 3. If you upload a track that has a lot of ambient noise in it be prepared for Suno to sizzle you with static. There's too many points for the AI to pick up, so you're just going to get a really messy output. 4. (This is a sidenote to uploading vocal tracks - the usual route of making electroswing manually is now POSSIBLE but takes like four covers minimum. Electroswing manually is - take a sample, usually jazz, then cover it with a house beat. I did this with Workman Swing, so it's now POSSIBLE. Just.... needs a lot of adjustment.) ==== 5. v4.5 is an interesting one, compared to the previous models, imo. It takes full sentence prompts now, and they can be EXTREMELY specific. I see a lot of experimentation in the future - long ago, when I first started, I used to play with using marketing copy in the styles box. This might work again, but INTERESTINGLY this time. Well, SSC7 is coming up soon, so I think I'll be sticking to my usual guns for it this time unlike last year's ambient liquid dnb/rock - electroswing + melancholy poetry (this is a preview of the vibe...except less techno). I should also maybe try out an alternating duet since I know how to do that cheaply now (see the previous notes for that). And maybe do a rap/sung combo? Geez. I've experimented a lot the last couple of months, that's cool I think. Well, see you guys next month (hopefully).

0/500