Creating Audio

Sogni Studio generates music. You can produce a full song with AI-written lyrics, a pure instrumental, or a short loop — all from a written prompt plus a handful of musical controls. The audio comes back in the gallery next to your images and videos and can be exported as standard audio files.

Audio runs on the Sogni Supernet using the ACE-Step model family.

#Two modes

#Song with Lyrics

A full song with AI-written lyrics structured into sections (verse, chorus, bridge, outro). Studio's LLM lyrics generator (the AI Lyrics Generator tool) produces:

• Section-based lyrics with vocal modifiers (e.g., "soft", "belted", "spoken").
• BPM, key, scale, time signature suggestions matched to the lyrics' mood.
• Duration for the full track.

You can edit any of the generated fields before rendering — rewrite a chorus line, change the key, shift the BPM. Or write the lyrics yourself and let the model handle the music.

#Instrumental

Music only — no vocals, no lyrics. Set BPM, key, scale, time signature, and a style prompt; ACE-Step composes around those constraints. Useful for:

• Background tracks for Sound to Video.
• Looping ambient beds.
• Reference tracks for your own composition work.

#Musical controls

Available across both modes:

• Prompt / Style — the descriptive direction ("dark synthwave with analog warmth and a driving bassline").
• Style Presets — around a hundred curated presets across genres, eras, moods, and production styles, plus a Custom field for your own style text. Stack with your prompt for finer direction.
• BPM — tempo. Auto-suggested per style, freely editable.
• Key & Scale — musical tonality (e.g., C minor, G Lydian).
• Time Signature — beats per measure. Defaults to 4/4; supports 3/4, 6/8, 5/4, 7/8, etc.
• Duration — total track length. Longer durations cost more Spark.

#How to use it

1. Switch Create mode to Audio.
2. Pick a mode — Song with Lyrics or Instrumental.
3. For lyrics: click Generate Lyrics to have the AI draft a full song, or write your own in the lyrics field.
4. Pick a style preset or write your own style prompt.
5. Tune BPM, key, scale, time signature, duration as needed.
6. Generate.

The job lands in the queue. Long tracks render in pieces and stream into the gallery as they're ready.

#Pairing audio with video

Studio's audio tracks plug straight into the video workflows:

#Screenwriter: cinematic prompt expansion

For video work that pairs with audio, Studio's chat panel includes a Screenwriter tool that takes a short scene description and expands it into a cinematic, LTX-tuned video prompt. Use it as the prompt input for video clips that match the mood of your generated music. See Chat & creative tools.

#Tips

• Start with a preset. The genre presets encode BPM, key, instrumentation, and production style as one click. Tune from there.
• Lock BPM before generating multi-track sets. If you want several tracks for the same project, lock the BPM and key so they layer cleanly.
• Iterate on lyrics, not music. Lyrics edits are cheap (you don't re-render the song until you generate); music re-rolls are full renders.
• Use 4/4 unless you have a reason not to. ACE-Step is strongest in 4/4; odd time signatures work but with more variance.

#See also