1. Voices – your own voice in the SUNO song
With the help of Voices, you can create your own voice profile from a recording, and then v5.5 uses this as the vocal voice of the generated song. You can also record live, upload an audio file between 15 seconds and 4 minutes, or choose a song from your own library.
The system checks whether it is really your voice by reading a random sentence. A clear vocal recording without music gives the best results. If there is background music, SUNO will try to separate the vocals.
2. Custom Models – own musical character
In the Pro and Premier packages, you can create a personalized v5.5 model by uploading at least six copyrighted songs. You can create up to three custom models and each one will remain private.
This is especially useful if you want a consistent Thirdvoice sound across multiple songs: your own model can better handle recurring instrumentation, rhythm, and mood. Not intended to copy other artists; only upload music for which you have the necessary rights.
3. My Taste - the magic wand already knows your taste
My Taste, available to all users, monitors the genres and moods you use frequently. If you press the Magic Wand icon in the Styles field, it creates a more personalized style description.
This is a quick starting point, but the final result is still worth fine-tuning by hand. The magic wand helps — it doesn't move into the studio for you.
4. Advanced Stem Separation – three types of lane separation
The June 2026 development offers three working modes:
According to SUNO, it does not simply "cut out" the frequencies, but regenerates the separate bands. This can give a clearer result, but it can also cause small differences compared to the original.
5. What about SUNO's business?
The company announced an investment round of more than $400 million in June at a $5.4 billion valuation. This isn't a new button on Create, but it does indicate that significant development money is behind the following music models.
What should you try first?
- Voices: create a short personal voice profile and try two versions with the same chorus.
- Custom Model: choose at least six of your own songs in a similar direction, not a completely mixed collection.
- Split from Mix: highlight the vocals or the guitar from a finished song, then check with headphones for any leftovers.
- Style Generator: use your own Our SUNO Style Generator 6-7 for your starting style tags.
What won't the new features automatically solve?
Your own voice profile also does not guarantee that every pitch, pronunciation and emotion will sound exactly as you imagined. Custom Model can give you a more uniform direction, but you can learn inconsistency from poorly selected teaching songs. Stems can be cleaner, but still have echoes, crosstalk or small regenerated differences.
Therefore, it is worth treating the new possibilities as a starting point. Listen to the individual tracks with headphones, compare multiple generations, and only use your own recordings that you know the rights to. Technology speeds up the work, but the final taste and quality control is still the responsibility of the creator.
After midnight in the SUNO studio
The prompt clearly describes it as “understated drums, delicate guitar, calm male vocals”. SUNO ponders, then enters with twelve drummers, three stadium guitars and a choir singing the line "silent night" like they're winning the World Cup in the process.
Gáspár takes down the strength of the drum, shortens the style, and regenerates it. Now the music is perfect, only the singer solves the word "in my heart" in fourteen syllables. Everything comes together on the third try - that's when it turns out that the chorus of the first version was better after all. This is not a mistake, this is the traditional initiation ceremony of the digital studio.