Suno just crossed a line that most AI music tools haven’t dared to touch: you can now clone your own voice inside the platform and use it directly in generated songs. The new version 5.5 makes that possible, and early results are genuinely impressive — not in a “technically interesting” way, but in a “this actually sounds like me” way.
The update also brings custom style models, a new My Taste feature that learns your musical preferences over time, and meaningful improvements to vocal expressiveness.
But the voice cloning is the headline, and it deserves a proper walkthrough — including the tricks that make the difference between a mediocre result and something you’d actually want to publish.
What’s New in Suno v5.5?
The update centers on three interconnected capabilities. First, voice cloning: you can now record, upload, or select audio of your own voice, and Suno will generate songs that sing in that voice.
Second, custom models: you can group a set of your own songs under a label and use that collection as a stylistic reference, teaching Suno a specific sonic direction rather than a single track.
Third, My Taste: a background analysis system that tracks your creative patterns — preferred styles, composition tendencies, recurring elements — so the platform can start anticipating your musical language over time.
💡 Key Insight
Voice cloning and custom models are designed to work together. Your voice captures how you sound. Your custom model captures how your music sounds. Combined, they give Suno enough of your creative identity to generate songs that feel like they belong in your catalog — not just generic AI output.
The old “Persona” feature is gone, replaced by a dedicated Voice section at the top of the interface. The shift isn’t just cosmetic — it signals that vocal identity is now a first-class feature, not an afterthought.
How to Clone Your Voice in Suno?
Step 1: Record or Upload Your Voice
Navigate to the Voice section and select Create Voice. You have three options: record directly in the browser, upload an existing audio file, or pull something from your Suno library. For recording, aim for at least 30 seconds — longer is better. Quality matters less than consistency: you want the recording to sound like you at your most natural, not like a one-take performance under pressure.
If you only have a fully produced track with no isolated vocal stem, that’s not a problem. Suno runs stem separation automatically when you upload a mixed file, so you don’t need to prepare an a cappella version in advance.
Step 2: Trim and Confirm
After recording, Suno lets you listen back and select the section you want to keep. Use this — the beginning and end of a take are rarely the strongest part. Keep the segment where you sound most consistently like yourself. Before proceeding, you’ll need to confirm that you have the rights to the voice you’re uploading. Once confirmed, click Use Voice.
Step 3: Identity Verification
This is the step that catches people off guard. Suno will display a short phrase and ask you to read or sing it aloud, in real time. The purpose is to confirm that the voice you uploaded actually belongs to you — a safeguard against people feeding in celebrity recordings or other third-party audio.
→ Practical tip for verification
If you recorded a singing voice, sing the verification phrase — don’t just read it. Verification appears to match the style and delivery of your original recording. Reading a phrase in a flat speaking voice when you uploaded a melodic take will likely fail the check.
Step 4: Set Your Singing Level Honestly
After verification, Suno asks you to classify your singing: beginner, intermediate, advanced, or professional. This isn’t about ego — it genuinely affects how the model processes your voice. A professional tag on a simple home recording will work against you. Match the label to what you actually recorded, and Suno will calibrate accordingly.
Step 5: Name and Save
Give your voice a descriptive name, optionally add style tags or a short description to help you identify it later, and save. Your voice will now appear under My Voices and can be selected for any future generation.
Getting the Best Results: The Slider Settings That Matter
Once your voice is saved, the generation settings determine how prominently it comes through in the output. Two sliders are critical: Style Influence and Audio Influence.
| Slider | Low setting | High setting |
|---|---|---|
| Audio Influence | Suno relies more on its own model — better overall audio quality | Closer to your recorded voice — more identity, potentially rougher |
| Weirdness | More predictable, stable results | Experimental, unpredictable variations |
A setting of around 40% on Audio Influence tends to be the practical sweet spot: enough to preserve your vocal character without sacrificing the quality Suno’s model brings. At 100%, the output leans too hard on your raw recording and quality suffers. Keep Weirdness at or near zero when you want consistent, usable results.
→ Think of Audio Influence as a blend knob
At 0%, you get Suno’s voice. At 100%, you get your raw recording through the model. At 40%, you get a version of you that Suno has cleaned up, tuned, and made production-ready. That middle range is where voice cloning becomes actually useful.
Custom Models: Teaching Suno Your Musical World
Voice cloning handles the “how you sound” side of the equation. Custom models handle the “how your music sounds” side. The feature lets you group a set of your own songs under a single label — say, a genre, a project name, or a specific vibe — and use that collection as a stylistic reference point for future generations.
The practical effect is consistency. Instead of describing a sonic direction from scratch every time, you point Suno at a set of songs that already embody that direction. Generations produced against that model will tend to share arrangement choices, tonal qualities, and structural patterns with your reference tracks.
The combination of voice cloning and custom models is where the real potential emerges. Your voice establishes who is singing. Your custom model establishes what kind of music they’re singing. Together, they give the platform enough context to generate tracks that feel like they belong together — the prerequisite for building a coherent album or EP with AI assistance.
💡 Key Insight
Custom models address one of AI music’s core weaknesses: inconsistency across generations. When every song sounds like it came from a different artist, you can’t build a project. By giving Suno a defined sonic world to work within, custom models make catalog-level consistency achievable for the first time.
My Taste: The Feature That Works in the Background
My Taste is the quieter addition to v5.5. It operates passively, analyzing the creative choices you make over time — the styles you return to, the structures you prefer, the sonic combinations you favor — and uses that profile to influence how Suno interprets your prompts. You don’t configure it directly; it builds from your activity. Think of it as a long-term creative memory that makes the platform more responsive to your instincts the longer you use it.
Who This Update Actually Changes Things For?
For casual users, voice cloning is a fun novelty. For anyone building a serious music project — whether that’s an artist developing a sound, a producer working with limited session budget, or a content creator who needs original music that sounds consistently like them — v5.5 moves Suno into genuinely useful territory.
The gap between “AI music that sounds AI-generated” and “AI music that sounds like me” has been the defining limitation of the category. This update doesn’t close that gap completely, but it closes it meaningfully. The verification step, the singing level classification, the blend sliders — these are the details of a system built with real creative workflow in mind, not just technical demonstration.
Frequently Asked Questions
Do I need a professionally recorded vocal to clone my voice?
No. A clean home recording works fine — even a simple melody hummed into a phone. The key is recording at least 30 seconds of consistent material and keeping the Weirdness slider low to let Suno’s model do the heavy lifting. The more naturally you sing during the recording, the more accurately the clone will represent your actual voice.
Can I upload a produced track instead of an isolated vocal?
Yes. Suno runs stem separation automatically when you upload a mixed audio file, so you don’t need to extract the vocal track yourself beforehand.
Why might my voice verification fail?
Verification compares your live delivery to the style of your uploaded recording. If you uploaded a singing voice but read the verification phrase in a flat speaking voice, the match will likely fail. Mirror the delivery style of your original recording — if you sang, sing the verification phrase.
What’s the difference between a voice clone and a custom model?
A voice clone captures how you sound as a singer. A custom model captures the musical style and direction of a set of your songs. They serve different purposes and are most powerful when used together: the voice makes the output sound like you, the custom model makes it sound like your music.
What Audio Influence percentage should I start with?
Around 40% is a reliable starting point. At that level, your vocal identity comes through clearly while Suno’s model handles the production quality. Adjust up if you want more character, down if you want cleaner output.
