Listen has released its biggest update yet to its AI music model with v5.5. Where previous updates mostly focused on improving fidelity and creating more natural tones, v5.5 is all about giving users more control. It includes three new features: Voices, My Taste and Custom Models.
In the release notes, Suno says that Voice is its most requested feature. It lets users train the vocal model on their own voice. They can upload ready-made tracks with clean acapellas, backing music, or sing directly into the mic on their phone or laptop. The clearer and higher quality the recording, the less data will be required. And to prevent someone from stealing another person’s voice, Listen also requires the user to speak a verification phrase. However, it may be possible to fool existing AI models of celebrity voices.
Once the voice feature is trained, users can sing an AI version of themselves over uploaded music or AI-generated output.
To further personalize the output, the Custom model allows users to train Suno on their music. Users must upload at least six tracks from their catalog and give the custom model a name. They will then be able to use this to guide v5.5 responses to signals.
Ultimately, my tastes learn tastes and preferences over time. You regularly come back for prompts about what genre, mood, and artist you are, and apply them while using the magic wand to auto-generate styles.
While My Taste will be available to all users, the Voice and Custom models are reserved for Pro and Premier customers.
<a href