· By Will Harken
2 Ways To Recreate Group Vocals With AI
AI can replicate voices, and that’s no secret. But what happens when you want to blend multiple singers? It can get tricky. This guide will help you navigate recreating those rich group vocals from your favorite tracks.
- Strategies for Handling Group Vocals
- Creating a Mono Group Vocal Model
- Creating Each Vocal Layer with a Solo Vocal Model
- Combining Both Strategies
- Mixing Tips for Group Vocals
Strategies for Handling Group Vocals
If you're using AI for group vocals, you essentially have two options.
Option 1: Create a mono group vocal model and then re-widen it. It's quick but may not provide the best quality.
Option 2: Create each layer with a solo model. This is slower but delivers a richer sound.
At the moment, AI models available to the public can't handle stereo group vocals as well as we’d like. But don’t worry—platforms like Udio can produce realistic stereo group vocals. Just remember, they can't convert existing vocals yet.
Creating a Mono Group Vocal Model
This strategy works when singers are singing in unison or at octaves. Take all the instances of those group vocals from the song to train your AI model. Just don’t expect radio-ready vocals.
This method is ideal for demos or music with a genuine feel.
Start by isolating the group vocals and letting the AI convert them into a group sound. Watch out for pitch problems or warbling; they can be tough to fix—even with tools like Melodyne.
The bright side? It saves you from having to recreate every tiny vocal detail from the original.
Creating Each Vocal Layer with a Solo Vocal Model
This approach means you’re re-engineering each vocal layer individually. It’s time-consuming and demands a keen ear for harmonies.
Imagine pulling together Freddie Mercury as your lead, Taylor Swift on the counter melody, and James Brown for backing. You’ll need to convert each voice separately and then mix them together.
While this method is labor-intensive, it yields top-notch results.
Combining Both Strategies
My go-to is often a mix of both strategies, especially when I have a lyrics swap to do (check out that service here). Layer a group AI vocal to give some background texture while bolstering it with high-quality solo harmonies for a fuller effect.
One can only hope that some tech wizard will create an automatic tool for this someday!
Mixing Tips for Group Vocals
When it comes to mixing, a compressor on your vocal bus can work wonders to “glue” those vocals together. Adding saturation can provide energy and mask any AI imperfections. I personally lean towards Nectar for this.
Check out tools like Nectar and Neutron. If you want personalized music, I'm here to help you elevate your projects.
Need a custom song? Let’s make it happen:
iZotope's Unmask is a fantastic tool for mixing. It cleverly ducks only those clashing frequencies instead of EQing the entire track—that's a game changer.
Explore iZotope plugins here. Keep your harmonies simple since AI often stumbles over complex ones. Training on one octave? That usually works. But multiple harmonies? Good luck with that.
Want to dive deeper into AI and music? Here are some articles you might like:
Blog Links
- Anyone Can "Sing" Now: AI Voice Cloning
- The AI Vocal Mixing Technique No One's Talking About
- How AI is Revolutionizing Music
Helpful Tools
- iZotope Plugins for AI Mixing and Mastering
- Synthesizer V - Digital AI Vocals That Sound Really Good
- AI Vocals: Kits.ai
- Isolate Vocals and Instrumentals with Lalal.ai
- Create Spoken Dialog with ElevenLabs
- Explore Weights.gg
- Check out Jammable (formerly Voicify)
Your journey into AI music-making is just beginning!