Create · sound · updated 22 October 2026

Music

A clip or a full song from a description, from pictures, or both. GPT Audio models can also say a short line out loud.

See the recipes Models for /music
  • 4music models
  • 10–200credits a track
  • 3mood pictures at most
  • 10sbetween /music commands

Entry I

Start here

Run /music with a prompt, up to 3 pictures (image1 to image3), or both. Leave model empty for Lyria 3 Clip (30s). The models are Lyria 3 Clip (30s): 100 credits; Lyria 3 Pro (full song): 200 credits; GPT Audio: 100 credits; GPT Audio Mini: 10 credits.

Credits are taken only when you press Generate. The button shows the price for the settings you picked, and your free 25 credits a day are spent before paid ones. A run that fails is refunded.

Entry II

Writing the prompt

Be specific

Genre, instruments, tempo in BPM, key, mood. Vague prompts give generic tracks.

Give it a shape

For a full song, name the sections: intro, verse, chorus, bridge, outro. Tags like [Verse] and [Chorus] help.

Lyrics apart from direction

Write your own lyrics as their own block, separate from the musical description. Write them in the language you want sung.

Instrumental means saying so

Add "instrumental only, no vocals" when you want no singing.

Pictures set the mood

Attached pictures are read as a mood, not as a script. Add a line saying what feel you want from them.

No real artists

Lyria's safety filters block a named singer's voice and copyrighted lyrics. Describe the style instead.

Entry III

Choosing a model

Lyria 3 Clip (30s) and Lyria 3 Pro (full song) make music and read pictures. The Clip model always makes about 30 seconds, so use it to test an idea; use the full-song model once the prompt is right.

GPT Audio and GPT Audio Mini speak or perform what the prompt says, in the "alloy" voice, and do not take pictures. For plain speech with a choice of voices, /speech is cheaper.

Entry IV

Enhance

Enhance has a small chat model (Claude Haiku 5.5, GPT-6 Luna, MiMo v2.6 Flash or DeepSeek V4.1 Flash 1017) rewrite your prompt for the model you picked, for +2 credits. You see both versions and choose Using: Yours or Using: Enhanced, and you can edit the enhanced one before generating. The rewrite targets the model you chose: a compact prompt for a clip, genre, instruments, tempo and structure for a full song.

Entry V

Every option

What you can type after /music, read from the command itself. You can run it again after 10 seconds.

OptionWhat it doesTakes
prompt Describe the music, or what GPT Audio should say text
model Lyria or GPT Audio textLyria 3 Clip (30s) (100 ◆), Lyria 3 Pro (full song) (200 ◆), GPT Audio (100 ◆), GPT Audio Mini (10 ◆)
image1 … image3 Reference image 1 file
hidden Only you can see this (default: false — post in channel) true / false

Entry VI

Recipes

Requests you can copy into Discord and change to fit.

A loop for a stream

Copy this

/music prompt: Lo-fi hip hop at 80 BPM in D minor, dusty vinyl crackle, mellow Rhodes chords, soft brushed drums, warm bass. Calm, late-night study mood. Instrumental only, no vocals.

Why it works. Genre, tempo, key, instruments, mood and "no vocals" in one paragraph. Good on the Clip model.

A full song with your lyrics

Copy this

/music prompt: Upbeat indie pop, 120 BPM, jangly guitars, handclaps, female vocal. [Verse] Monday rain on the window, coffee going cold [Chorus] But we're still here, still here, singing all the way home model: (full song)

Why it works. Direction first, then the lyrics in tagged sections.

Music for a picture

Copy this

/music prompt: Music for this place: slow, spacious, a little lonely. image1: (a misty mountain photo)

Why it works. The picture gives the mood; one line says what to take from it.

Entry VII

Tips

  1. I

    Test a prompt on the Clip model, then make the full song.

  2. II

    Name two or three instruments. A long list blurs together.

  3. III

    Say "no vocals" if you want none; otherwise a model may sing.

Entry VIII

Limits

  • One music job at a time, 10 seconds between commands, and a 5 minute limit per track.

  • Up to 3 pictures, and only on models that read them.

  • Lyria refuses named artists' voices and copyrighted lyrics.

Entry IX

Questions

How long is a track?

The Clip model makes about 30 seconds. The full-song model decides the length from your prompt; say how long you want it.

Do I need a prompt?

No. Pictures alone are enough; music is composed to match them.

Can I edit a track?

Each run is a new track. Change the prompt and generate again.

/speechSpeech /imageImage /wikiWiki

Where the advice comes from

Prompting advice is from these free public guides, matched to what ALMA lets you set. Prices and limits are read from the bot itself, as of 22 October 2026.

Try it

Type /music in any server with ALMA, or in a direct message to the bot.

All guides

Something here wrong, or missing? Tell us.