Music: songs, covers and stems

How to make a song from a description or your own lyrics, continue or rework a track, split a recording into stems and save a voice for new songs. The music is made by Suno, an AI that writes and sings songs. It runs in the web studio.

Updated:

A song from a description

This is the easiest way to get a finished song. You say in plain words what it should be like, and Suno writes the lyrics and the music and sings it. For example: “a cheerful pop song for my friend’s birthday, female voice”.

  1. In the studio, open Music, go to the Create tab and pick the Song operation.
  2. In the Describe the song field, write the genre, the mood, what the song is about and what voice you want.
  3. Choose the Suno version: V6 — main, V6 Mini — fast, V6 Wild — bold.
  4. Press Start. The song appears in the Library in two versions, so you can keep the one you like better.
Example description
A summer indie-pop song about a drive to the sea in an old car, light female vocals, guitar and synths, a chorus you want to sing along to
  • The description can be up to 3,000 characters.
  • Want music without singing? Turn on Instrumental. You get an instrumental: just the music, no voice.
  • While the song is being made, the bar above the library shows the progress: “Writing lyrics…”, then “First track is ready, finishing the second…”.
  • You can play a finished track right in its card and save it with Download MP3.
Stained glass: a vintage car driving along a coast road by the sea

Your own lyrics and style

Use this mode when you already have words or know exactly how the song should sound. For example, paste your own birthday poem and ask for “acoustic guitar, warm male voice, slow”.

In the Song operation, turn on Custom lyrics and style. The single description field is replaced by separate ones:

  • Title — up to 80 characters.
  • Style — how the song should sound: genre, instruments, tempo and mood, separated by commas. Up to 1,000 characters. Example: “synthwave, 110 bpm, female vocal, dreamy”; 110 bpm is the tempo, 110 beats per minute.
  • Lyrics — up to 5,000 characters. You can label the parts of the song in square brackets: [Verse], [Chorus], and [Bridge], a short part that sounds different from the verse and the chorus.
  • Exclude styles — styles the song must not have, such as “rap, electronic”. Up to 200 characters.
  • Vocal gender — male or female.
  • Persona or voice — a saved singing manner or your own voice. How to make them is explained in “Personas and your own voice” below.
  • Duration, s — from 10 to 360 seconds.

Three more fields are optional. Style weight (0 to 1) is how closely to follow the Style field. Weirdness (0 to 1) is how unusual the sound may get. Variety goes from 0 to 4.

No lyrics yet? The Create tab has two helpers:

  • Lyrics — write a short idea in What the song is about (up to 200 characters) and get two or three sets of lyrics. The Copy button copies the one you like so you can paste it into your song.
  • Boost style — type a couple of words about the style (up to 300 characters), such as “sad jazz”, and get a detailed description for the Style field.

What you can do with a track

If one part of a song doesn’t work, you don’t have to start over. You can continue the song, rewrite a piece of it, have it sung in another genre, or put vocals on the music. For example, turn a pop song into a jazz version, or add another verse to a short track.

In the Library, press Actions on a track and choose an operation. A form opens with the track’s title, style and lyrics already filled in. Change what you need and press Start.

  • Extend — Suno writes a continuation. In Continue from second, set where it should start; in Lyrics for the continuation you can add words for the new part. You get two versions.
  • Replace section — redoes one piece of the song and leaves the rest as it was. Handy when a single line or the chorus didn’t come out right. Set the section start and end in seconds, the New lyrics for the section and the Full lyrics after the change, which is the whole text with the new words in place. The section must be at least 10 seconds long and no longer than half of the track.
  • Cover in another style — a cover is the same song performed in a new way: the melody stays, the genre and sound change. Type a New style, such as “jazz, saxophone”. You get two versions.
  • Add vocals — Suno sings on top of the track. It works best on an instrumental, a track without a voice. Set the Vocal style and, if you like, the lyrics.
  • Add instrumental — the other way round: Suno writes music to go with the track’s vocals. Set the Instrumental style.
  • New cover images — two new images for the track’s cover. You can do this once per generation.
  • The same menu has stem splitting, export and saving a persona. They are covered below.

A mashup blends two songs into one. Above the library, press Mashup of two tracks, tick two tracks and press Next. You get two versions.

Extend, Replace section, New cover images, stem splitting, export and saving a persona only work on tracks made in the studio, within 14 days of their creation. A track from your own file can be used for a cover, added vocals, an added instrumental and a mashup.

A frog trio on a lily pad, jazzy mood

Your own file

You can do the same things with your own recording. Hum a tune into your phone and Suno can turn it into a rock version. Record a guitar part and Suno can sing over it.

Open the Your file tab, pick an operation, upload an audio file in MP3, WAV, M4A, OGG, FLAC or AAC and press Start. The maximum file size is shown under the file picker.

  • Cover your track — your recording in a new style, keeping the melody. The recording can be up to 8 minutes long.
  • Extend your track — Suno continues your recording from the second you choose. The recording can be up to 8 minutes long.
  • Vocals for your instrumental — upload music without a voice and Suno sings over it.
  • Instrumental for your vocals — upload a voice and Suno writes music for it.
  • Split your track — splits the recording into stems, see the next section. The file can be up to 20 MB. Afterwards the recording appears in the Library, with the stems in its card.

Suno does not accept published recordings. If an uploaded file matches an existing work, such as a song by a well-known artist, the operation is rejected and the tokens go back to your balance. Upload your own recordings or music you have the rights to.

Stems, MIDI and export

“Split into stems” takes a finished song apart into separate parts: the voice on its own, the drums on their own, the bass on its own and so on. Use it to make a karaoke backing track, pull out a clean vocal for a remix, or take just the drums or bass from a track for your own arrangement. You can split a track from your library or your own file — on the “Your file” tab, “Split your track”. There are three modes:

  1. Vocals and instrumental — two stems: the voice and all the music without it.
  2. All instruments — up to 12 stems: vocals, backing vocals, drums, bass, guitar, keys and more.
  3. One chosen stem — for example, only the drums or only the saxophone; it comes with the track minus that part.

Each stem can be played and downloaded on its own.

For a library track, press Actions → Split into stems, choose the mode in What to extract and press Start. The price depends on the mode. The stems appear in the track’s card.

MIDI is not sound but notes: which note plays, when, and for how long. You open a MIDI file in music software to replay a part on another instrument or fix the notes. After splitting, the track’s card has a MIDI from stems button next to the stems. Save the result with Download MIDI.

Export gives you files of the track to use in other apps. You order them from the Actions menu too:

  • Lossless WAV — the same track as a WAV file. MP3 throws away some of the sound to keep files small; WAV doesn’t. That is why WAV is used for mixing (combining parts into one track) and mastering (the final polish of the sound).
  • Timed lyrics — the lyrics with the time each word is sung. Useful for karaoke and subtitles. You can read them in the track’s card or save them as an LRC file, a lyrics format for karaoke.
  • Music video — an MP4 clip with the cover and a sound visualiser; you can add a Signature on the cover.

Export and Split into stems in the Actions menu only work on tracks made in the studio that are no more than 14 days old.

Kites flying over green hills

Personas and your own voice

A persona is a saved singing manner: the voice, delivery and style from a track you liked. Pick it for a new song and Suno sings it the same way, as if it were the same performer. That lets you make several songs by “one artist”, such as an album.

To save a persona:

  1. In the Library, press Actions → Save as persona on a track.
  2. Choose the part where the singing is clearest: Fragment start and Fragment end, in seconds. It must be 10 to 30 seconds long.
  3. Name the persona and describe it briefly in Description, for example “raspy male voice, blues”.
  4. Press Start. The persona appears on the Voice and personas tab, under Your personas and voices.

Your own voice works the same way, but with your real voice: new songs are sung in your voice. You need a clean recording of yourself singing without music. The Your voice wizard on the Voice and personas tab has three steps:

  1. Voice sample. Upload the voice recording, set the start and end of the part to analyse, choose the Phrase language and press Your voice: sample.
  2. Verification phrase. The studio shows a short phrase. Sing or read it in your own voice and upload the recording. This is how the service checks that the sample is your voice. If the phrase is awkward, press Another verification phrase. Then name the voice, choose the Singer skill level and press Your voice: create.
  3. Done. The voice appears under Your personas and voices.

You pick a persona or a voice in the Persona or voice field. It is in Song with Custom lyrics and style turned on, in Extend, Replace section, Cover in another style and Mashup of two tracks, and in Cover your track and Extend your track.

  • A persona can only be saved from a track made in the studio, within 14 days of its creation.
  • Verification phrase languages: Russian, English, Spanish, French, German, Portuguese, Japanese, Korean, Hindi or Chinese.
  • You can delete a persona or a voice under Your personas and voices; tracks already made with it stay.
A microphone with flowers growing out of it

Prices and refunds

Every music action is a separate operation with its own price in tokens. You only pay for what you run: a song and splitting it into stems, for example, are two different operations.

You see the price before you start: next to the operation name, in the Actions menu and on the Start button. For Split into stems it depends on the mode. The full list is on the music page. Song, Extend, Cover in another style and a mashup return two versions for one price.

If an operation fails, the tokens go back to your balance automatically. That happens when the generation service returns an error or runs out of time, the request does not pass the filter, or an uploaded recording matches a released one. The studio tells you the reason.

How to make a song

  1. Open Music. In the web studio, open Music, the Create tab and the Song operation.
  2. Describe the song. In Describe the song, write the genre, mood, topic and the voice you want. Or turn on Custom lyrics and style and fill in the title, style and lyrics.
  3. Pick a Suno version. V6 — main, V6 Mini — fast, V6 Wild — bold.
  4. Start. Press Start — the price is on the button. Two versions of the song appear in the Library.
  5. Polish the track. In the track’s card, press Actions to extend the song, replace a section, make a cover or split it into stems.