The whole song in one pass
Describe a song and Suno v5.5 writes the words, sings them, and arranges the parts around them. It is the model Suno released alongside Voices, and the default version on every generation endpoint here.
One Last Kiss
orchestral, dream pop
Generated on this site with Suno v5.5 from a one-line description. Nothing was edited afterwards.
The composer below is the one inside the product, running on this page. Describe a song and it comes back in about a minute.
Warm pop rock / soul, 88 BPM, B-flat major. Intimate reflection swells into an empowering anthem of self-worth, led by a seasoned female alto with a velvety light rasp, analog-warm guitars, steady drums, soulful keys, reassuring hooks, and wide cinematic choruses.
Upbeat funk / nu-disco, 112 BPM, E-flat major. Confident, strutting and celebratory, with a smooth soulful male tenor, playful staccato phrasing, falsetto flips, snappy drums, elastic bass, rhythmic guitar, glossy keys, and a wide dance-floor mix.
Six songs generated here. Cover art and all, straight out of the model.
Suno v5.5 is Suno’s most expressive music model. Full songs with vocals, lyrics, and arrangement from a single description — then split, extended, and rewritten without leaving the track.
Suno released v5.5 in March 2026 and called it its most expressive model yet, launching it alongside Voices — the most requested feature in its community — along with custom models tuned to a creator’s own catalogue. In this music generator it is the default version behind every generation, and the only version some operations will accept.
Suno calls the human voice the oldest instrument we have, and Voices was the most requested feature in its community. Save the singer from a song you made — with its style, or without — and it becomes an option on every future generation. Suno verifies a captured voice by asking you to speak a random phrase and matching it against the singing in your audio, and keeps the result private to your account.
Every Suno v5.5 track splits into twelve separate audio files. These are the real output of a split run on the song above, and each shape below is that part’s own waveform.
The company's own launch films for the release.
The announcement
Meet Suno v5.5: more expressive, more you.
Why it is different
Suno walks through what changed in the model.
Suno’s own walkthrough of the Suno v5.5 release, played from YouTube.
What changed
Every prompt returns two finished songs, so the first question is never “is this good enough” but “which of these two is closer”. That is a different question, and a faster one. Whichever you keep becomes the source every later edit points at — the extension, the crop, the cover and the twelve stems all address that take rather than re-running the prompt.
Describe the song
A sentence is enough. Name the mood, the instruments, and the tempo, and let the model write the words — or write them yourself and mark the sections.
Pick the take
Two finished songs come back, usually inside two minutes. Play both and keep the one that reads right.
Work on the one you kept
Extend it, replace a section, save its singer, split it into parts, or take it away as MP3, WAV, MIDI, or a music video.
One pass produces the lyric, the performance, and the parts behind it. There is no separate lyric step to paste from, and no instrumental to sing over afterwards.
Each generation returns two complete songs from the same prompt rather than one. Keep the reading you prefer and carry it into every later edit.
Suno shipped v5.5 alongside Voices, the most requested feature in its community. Save a singer from a song you made and reuse that character on the next one.
A finished track splits into vocals, backing vocals, drums, bass, guitar, keyboard, percussion, strings, synth, FX, brass, and woodwinds — each as its own file.
Extend from any second, replace a section, crop, remove a passage, fade, or change tempo. Every edit references the original take rather than re-rendering from the prompt.
Take the master as MP3 or WAV, the notes as MIDI, or a shareable music video — from the same menu as the edits.
Styles are a list, not a sentence
The styles field takes comma-separated tags — “orchestral, dream pop, 110 BPM” — and reads them as a set of constraints. Prose belongs in the description instead.
Mark the sections when you write the words
Wrap parts in [Verse], [Chorus], [Bridge], and [Outro]. The model uses the markers to shape the arrangement, not just the lyric.
Say what you do not want
Exclude styles explicitly rather than implying them. “No harsh distortion” in the description is weaker than putting it in the exclude field.
A slow orchestral dream pop song about leaving a city at dawn. Strings under a soft male lead, brushed drums, 96 BPM, warm tape hiss.
Five versions are available in this Suno music generator. Suno v5.5 is the default everywhere, and the only one some operations accept.
Editing operations — crop, remove section, fades, tempo, stem separation, and saving a voice — have no version dimension at all. They act on the file the song already is.
Hear an arrangement of an idea before committing a day to it. Two takes per prompt means the first thing you hear is already a comparison, and the stems come out if a part is worth keeping.
Score a cut without licence clearance. Generate to a mood, crop to the exact length the edit needs, and fade the ending into the next scene.
Produce loops and one-shots alongside full tracks, at a set tempo and key, then split out the layers a mix needs to fade in and out at runtime.
Save a voice from a track you like and keep the same singer across a release. Covers reinterpret an existing song in another style without rewriting the words.
Suno v5.5 is the music model Suno released on 26 March 2026, described by the company as its most expressive to date. It generates a complete song — lyrics, vocal performance, and arrangement — from a written description, and it is the default version used by every music generation on this site.
Suno positioned Suno v5.5 around expressiveness and personal identity rather than raw audio quality, launching it alongside Voices, Custom Models, and My Taste. Practically, it is the default version here and the only one that some operations accept: adding a stem to an existing track requires v5.5, and earlier versions stall on that endpoint.
Typically between thirty seconds and two minutes. Every generation returns two complete takes of the same prompt, so you are choosing between finished songs rather than waiting for a second attempt.
Yes. Switch the composer to Write and supply the words yourself, using [Verse], [Chorus], [Bridge], and [Outro] to mark the structure. Leave it on Prompt and the model writes the lyric from your description instead.
Yes. Any prompt can be rendered without vocals, and instrumental tracks can still be extended, cropped, covered, and split into parts afterwards.
Stems are the separated parts of a finished song. Splitting a Suno v5.5 track returns twelve files: vocals, backing vocals, drums, bass, guitar, keyboard, percussion, strings, synth, FX, brass, and woodwinds.
Yes, and the edits address the original take rather than re-running the prompt. You can extend from any second, replace or remove a section, crop, fade the start or ending, change the tempo, or reinterpret the whole thing as a cover in another style.
MP3 and WAV for the audio, MIDI for the notes, and a shareable music video. Stems download as individual audio files. All of them are produced from the track menu.
Two finished takes from one description, and every edit in the same place.
