If you put in lyrics and maybe extend the prompts a little bit more (like "energetic vibe, strong bass line, dynamic drummin") you're going to get a pretty coherent output. The longer the input audio file is, the better it can mimic it. If you're going fully into it, you can even get it to cover any specific song you like. But you have to give it more input, otherwise it will fill the blanks itself and you gonna get VERY random outputs.
Udio and Suno are really rough with this. I think they need to implement a function like Kits and allow you to pick a vocalist for the generation. This could be a great start if you have a vocal idea though, but just don't know how to integrate it.
If you want the song to start with your part, just remember to click the "Get whole song" this is an extended, starting from where your clip ends, and yeah writing lyrics yourself would help the gibberish problem... allthough that can happen some times in either way
The emergence of Suno has greatly helped those who have the desire to create but have always struggled with the process. Suno allows people to understand how a small spark of inspiration can develop into a complete song and how this process is specifically presented. As a professional guitarist, I really enjoy using Suno as a creative tool and am using it to assist with my daily creations. I really like your videos and hope your work continues to improve!
the reason why it did gibberish is because you didn't put any lyrics in the lyrics tab, it's like setting the style to rap and catchy lyrics and it being gibberish but rhyming 😭 edit: renaming it was in edit details 😭x2
I like how it takes your voice/vocals and modulates and mixes up melodies using your voice .. this is the way to use ai tools by using your own voice or unfinished work to finish your ideas up .. better than a person just using prompts without their own voice/work
Hey, hi, great videos!! So, have you got a chance to play with the meta-tags (the prompts that we can provide within lyrics in [ ] like [chorus], [verse], etc) to see how can we prompt Suno to sing a verse in male voice and another in female voice?
I just like to write songs and since I don’t play instruments It takes me harder to come up with melody to record vocal demos to show the lyrics. Does this clone your voice to reuse for different lyrics and help you do melodies?
I tried doing that on my iPad right now, Suno said you can only upload only one file (on my iPad) and the singer’s voice was nowhere near to imitating my voice, I think.
That's not using autotune at all, it just take your audio as reference and make one from scratch. What you call autotune is a lack of training on the vocals that do this.
Yes. I played a rhythm with a wooden drum and Suno use my recording and also the hole music in that rhythm. I tried different styles.some came out great!
Why are you surprised it was gibberish when you didn't give it any lyrics to sing? 😂 Did you expect it to write its own lyrics? Custom Mode doesn't do that, it expects you to provide the words. Or at least to click the "Make Random Lyrics" button before submission 😁 A lot of people have asked for Suno to allow downloading of stems, but that's just not how the model works. The model is trained on, and generates, one raw audio sample at a time. So it generates all the instrumentation and vocals and every bit of audio in the same one pass. The only way to get stems would be to have a totally different model trained to generate stems -- so, not a Suno v4 but a brand new model architecture -- or to use a separate AI after generation to post-process it and split the stems (which won't sound as clean as actually having the stems would be).