← Back to HookGenius

Suno v5.5: Voices, Custom Models & My Taste

Published: March 27, 2026

Suno v5.5 launched March 26, 2026. It's not a new audio engine — it's a personalization layer on top of v5. Three features: Voices (clone your own voice), Custom Models (train Suno on your music), and My Taste (automatic preference learning). Your prompts, metatags, and style tags all work exactly like v5.

New to Suno? Start with our complete prompting guide. Already know v5? This page covers only what's new.

v5 vs v5.5: What Actually Changed

v5.5 is an additive update. Everything from v5 carries forward. Here's what's new:

Feature v5 v5.5
Voice Cloning Not available Full support — clone your own voice
Custom Models Not available Train on your catalog — up to 3 models
Personalization Manual only My Taste — learns your preferences automatically
Audio Quality Studio-grade Studio-grade (same engine)
Style Prompt Limit 1,000 characters 1,000 characters (unchanged)
Lyrics Limit 5,000 characters 5,000 characters (unchanged)
Metatags Full support Full support (unchanged)
Negative Prompting Full support Full support (unchanged)
Studio / Stems Full support Full support (unchanged)
Expressiveness High Higher — more responsive to nuanced descriptors

Key takeaway: If you already have prompts that work well in v5, they'll work in v5.5 — potentially even better. The new features layer on top of prompting, they don't replace it.

Voices: Clone Your Own Voice

This is the headline feature. Upload or record your vocals and Suno will use your actual voice on generated tracks.

How Voices Work

Getting the Best Voice Clone

  1. Clean recording matters. Minimal background noise. No heavy reverb. The cleaner the input, the better the clone.
  2. Provide range. If possible, include both your lower and higher register. Monotone input = monotone output.
  3. Natural delivery. Suno captures your natural vocal tone, not a performance. Sing or speak normally.
  4. Multiple samples help. One clip works. Three clips across different moods works better.

Availability: Voices requires a Pro ($10/mo) or Premier ($30/mo) subscription. Free tier users cannot create voice clones.

Pairing Voices with Detailed Prompts

Your cloned voice is the singer. The prompt is the song. The more specific your prompt, the better the voice sits in the mix.

[Verse] Under the streetlight haze, counting cracks in the road Every scar on my hands tells a story untold [Chorus] I built this from nothing — brick by brick, late by late The world said wait your turn, I said I'll make my own fate STYLE PROMPT: indie rock, raw production, 128 BPM, driving drums, overdriven guitars, emotional male vocals, gritty, no autotune, no polish, garage recording vibe

Pro Tip: When using a cloned Voice, you can drop gender descriptors from your style prompt (like "male vocals" or "female vocals"). Suno already knows the voice. Use that character space for more production detail instead.

Custom Models: Train Suno on Your Sound

Upload your original music and Suno fine-tunes v5.5 to learn your production DNA — how you mix, what instruments you favor, your signature sound.

How Custom Models Work

When to Use Custom Models

Custom Models + Prompts: The Power Combo

Custom Models set the baseline production style. Your prompt steers the specific song. Think of it like a producer (Custom Model) working with a songwriter (your prompt).

Without Custom Model: "lo-fi hip hop, warm vinyl crackle, jazzy chords, mellow, 85 BPM, boom bap drums" → Generic lo-fi output With Custom Model (trained on your beats): "upbeat summer track, 110 BPM, playful melody" → Sounds like YOUR lo-fi style, but faster and brighter

Pro Tip: Custom Models work best when your training catalog has a consistent style. If you upload 5 lo-fi tracks and 5 metal tracks, the model gets confused. Train separate models for separate styles.

My Taste: Passive Personalization

My Taste is the simplest new feature — and the one available to everyone, including free users.

How My Taste Works

Suno watches what you generate, what you like, what you skip. Over time it builds a preference profile:

This shapes default suggestions, recommended styles, and subtle generation tendencies. It's not a switch you flip — it evolves as you use Suno.

Note: My Taste influences defaults, not your explicit prompts. A detailed style prompt always overrides My Taste preferences. This is why writing specific prompts still matters in v5.5.

v5.5 Prompting Strategy

v5.5 didn't change prompt syntax, but its improved expressiveness means your prompts land harder. Here's how to take advantage.

More Nuance Gets Rewarded

v5.5 is better at interpreting subtle descriptors. Where v5 might ignore "slightly detuned vintage keys," v5.5 actually delivers it.

v5 prompt (still works): melodic trap, dark, 808s, hi-hats, moody vocals v5.5-optimized (same format, more detail): melodic trap, 145 BPM, deep sub 808s, glitchy hi-hat rolls, pitched vocal chops, dark atmospheric pads, reverb-drenched ad-libs, minor key, no bright synths

Stack Your Personalization

The real power of v5.5 is layering: detailed prompt + Voice + Custom Model. Each layer adds specificity.

The v5.5 Stack

Layer 1: My Taste — shapes defaults based on your history
Layer 2: Custom Model — sets production DNA and sonic identity
Layer 3: Voice — applies your actual vocal tone
Layer 4: Prompt — steers the specific song (lyrics + style tags)

The prompt is still the most important layer. It's the one you control song by song. Detailed prompts from HookGenius give v5.5's personalization engine something precise to work with.

What to Change in Your Prompts (and What Not To)

Example: Same Song, Three Approaches

Free user (My Taste active): dark r&b, 80 BPM, sparse production, whispered vocals, reverb-heavy, late-night, no drums in verse, minimal Pro user (+ Custom Model): dark r&b, 80 BPM, whispered vocals, reverb-heavy, late-night, stripped verse builds to heavy chorus [Custom Model handles the production DNA — free up prompt space for arrangement] Premier user (+ Custom Model + Voice): dark r&b, 80 BPM, reverb-heavy, late-night, stripped verse builds to heavy chorus, close-mic intimacy [Voice handles vocals, Custom Model handles production — prompt focuses on song architecture]

Metatags in v5.5

All metatags from v5 work identically in v5.5. No changes to syntax, no new tags specific to v5.5. The full list:

Structure Tags

Delivery Tags

For the complete 300+ tag reference, see our Suno metatags guide.

Pro Tip: v5.5's Voices feature pairs especially well with delivery tags. Clone your voice, then use [Whispered] or [Belted] to get YOUR voice in different performance styles. The model respects these even with custom voices.

Who Gets What in v5.5

Free Tier

v5.5 audio quality, My Taste personalization, limited daily credits. Same prompt system as paid users.

Pro — $10/month (2,500 credits)

Everything free tier gets, plus: Voices (voice cloning), Custom Models (up to 3), Studio access, commercial use rights.

Premier — $30/month (10,000 credits)

Everything Pro includes, plus: priority generation queue, highest audio quality, early access to new features.

Best Practices for v5.5

  1. Start with your prompt. Voices and Custom Models enhance output, but the prompt is still the foundation. A vague prompt with a great Voice still produces a vague song.
  2. Train Custom Models on consistent catalogs. Don't mix genres in one model. One model per style.
  3. Clean recordings for Voices. Background noise, heavy effects, and low-quality audio all degrade the clone.
  4. Let My Taste settle in. It needs usage data. After 20-30 generations, you'll start noticing it shaping defaults.
  5. Use freed-up prompt space wisely. If your Voice handles vocals and your Model handles production, use that character budget for arrangement, energy dynamics, and song-specific detail.
  6. Generate multiple versions. v5.5 is more varied than v5. Generate 3-5 versions and pick the standout.

Frequently Asked Questions

What's new in Suno v5.5 compared to v5?

Three personalization features: Voices (clone your own voice), Custom Models (train on your music catalog), and My Taste (automatic preference learning). The core audio engine, prompt format, and metatag system are unchanged from v5.

Does v5.5 change how prompts work?

No. Same 1,000-character style prompt, same 5,000-character lyrics, same metatags. Your v5 prompts work identically in v5.5 — and may actually sound better due to improved expressiveness.

Can I use someone else's voice?

No. Anti-deepfake verification requires you to speak a random phrase live. Voices are private and locked to your account.

How many Custom Models can I create?

Up to 3 on Pro and Premier plans. Free users cannot create Custom Models.

Do my old prompts still work?

Yes. All v5 prompts, metatags, negative prompts, and style tags work identically in v5.5. Nothing was removed or changed.

Is My Taste worth it?

It's passive — you don't configure it. Just keep using Suno and it learns. After a few weeks of usage, you'll notice more relevant default suggestions. Detailed prompts always override My Taste preferences.

Should I update my HookGenius prompts for v5.5?

No changes needed. HookGenius prompts are already v5-optimized with detailed style tags, metatags, negative prompting, and BPM. These work perfectly in v5.5. If you're using Voices, you can optionally drop gender descriptors to free up prompt space.

Can I combine Voices + Custom Models?

Yes. This is the most powerful combination in v5.5: your voice singing over production that matches your established style, directed by a detailed prompt. The three layers stack.

Generate v5.5-Ready Prompts Instantly

HookGenius creates detailed lyrics and style prompts that unlock v5.5's full expressiveness. 50+ genres, negative prompting, BPM control, vocal timbre, energy dynamics — all optimized for Suno.

Try HookGenius Free

Suno Guides

Level Up Your Music