Music Gen Usage

AI Music & Sound Effects

• compose_music — Compose an original music track from a text description (ElevenLabs Music). Returns a workspace MP3 path and shows an inline player.
• sound_effect — Generate a sound effect from a text description (ElevenLabs SFX).

Guidelines

• Describe music concretely: genre, mood, instrumentation, tempo, and energy arc (e.g. "uplifting electronic track, driving synths, punchy four-on-the-floor beat, builds to an energetic drop"). Vague prompts ("nice music") produce generic results.
• Set instrumental: true for video soundtracks and background music — otherwise the model may add vocals.
• Match duration_seconds to the actual use: don't compose a 3-minute track to score a 30-second clip. Longer tracks cost more credits.
• For sound effects, describe the sound's character and envelope ("soft UI confirmation chime, glassy, short decay"). Omit duration_seconds to let the model pick a natural length; use loop: true for ambience beds.
• Generated audio is auto-saved to the workspace under generated-assets/ (a bare filename passed via save_path also lands there). Use a relative path with a subfolder (e.g. "projects/launch/theme.mp3") to organize into your own folder.
• The saved MP3 is web-safe: play it with display_audio, or feed it to video-tools as a soundtrack.
• Generation is not instant — a long track can take a minute or more; this is normal.