Documentation — Coming Soon

AI Music Generation

Create music and sound effects from text descriptions — available in a future update

AI Generation Basics

Generate Music from Text

Stori uses Meta's AudioCraft models to generate music and sound effects from text descriptions:

  • ○ Music Generation: Generate music clips in any genre
  • ○ Sound Effects: Create sound effects and environmental audio
  • ○ 100% Local: All AI will run on your Mac
  • ○ Template System: Pre-built prompts for quick generation
  • ○ Prompt Enhancement: AI improves your descriptions
  • ○ Timeline Integration: Generated audio appears instantly
  • ○ No API Fees: Unlimited generations, no cloud costs
🤖

AI Generation modal with prompt templates

🎵 MusicGen vs 🎧 AudioGen

🎵

MusicGen

Music Generation

Generate musical compositions with melody, harmony, rhythm, and instruments. Perfect for creating songs, loops, and musical ideas.

  • Full musical compositions
  • Multiple genres (EDM, hip hop, rock, jazz, etc.)
  • Tempo and key control
  • Instrument specification
  • Mood and energy settings
  • Melody conditioning support
"High-energy EDM track at 128 BPM in A minor with punchy synths and driving bass"
🎧

AudioGen

Sound Effects

Generate sound effects, foley, ambience, and textures. Ideal for podcasts, videos, games, and adding atmosphere to music.

  • Sound effects (footsteps, doors, etc.)
  • Environmental ambience
  • Foley and textures
  • Nature sounds
  • Mechanical/industrial sounds
  • Atmospheric layers
"Rain on a tin roof with distant thunder and occasional wind gusts"
⚡

Quick generation workflow

Quick Start

Generate Your First Track

Simple workflow to create AI-generated music:

  • 1 Click Generate: Click "Generate" button on track or ⌘⇧G
  • 2 Choose Type: Select MusicGen (music) or AudioGen (effects)
  • 3 Enter Prompt: Describe what you want to create
  • 4 Enhance (Optional): Click "Enhance Prompt" for better results
  • 5 Set Duration: Choose 5-30 seconds
  • 6 Generate: Click "Generate" button and wait ~30-60 seconds
  • 7 Listen: Audio appears on timeline, press Space to play

📋 Prompt Templates

Pre-built templates help you generate professional results quickly. Choose a template and customize it:

🎹 Electronic Music

  • House: "Upbeat house track at 128 BPM with four-on-the-floor kick and disco strings"
  • Techno: "Dark minimal techno at 130 BPM with industrial percussion"
  • Trance: "Uplifting trance at 138 BPM with epic synth pads"
  • Dubstep: "Heavy dubstep at 140 BPM with wobble bass"

🎤 Hip Hop & R&B

  • Boom Bap: "Classic 90s hip hop beat at 90 BPM with vinyl crackle and jazz samples"
  • Trap: "Modern trap beat at 140 BPM with 808 bass and hi-hat rolls"
  • Lo-Fi: "Chill lo-fi hip hop at 85 BPM with mellow piano and vinyl noise"
  • R&B: "Smooth R&B groove at 75 BPM with electric piano"

🎸 Rock & Metal

  • Rock: "Classic rock riff at 120 BPM with distorted guitar and driving drums"
  • Metal: "Heavy metal at 180 BPM with double bass drums and palm-muted guitars"
  • Punk: "Fast punk rock at 160 BPM with power chords"
  • Indie: "Indie rock at 110 BPM with jangly guitars"

🎺 Jazz & Blues

  • Jazz: "Smooth jazz at 120 BPM with saxophone and walking bass"
  • Blues: "12-bar blues at 80 BPM with blues guitar"
  • Bebop: "Fast bebop at 200 BPM with intricate melody"
  • Fusion: "Jazz fusion at 140 BPM with electric piano"

🌙 Ambient & Cinematic

  • Ambient: "Ethereal ambient soundscape with evolving pads and subtle textures"
  • Cinematic: "Epic cinematic orchestral piece with strings and brass"
  • Drone: "Dark drone atmosphere with low rumble and distant echoes"
  • New Age: "Peaceful new age music with piano and nature sounds"

🎧 Sound Effects

  • Nature: "Forest ambience with birds chirping and rustling leaves"
  • Weather: "Heavy rainstorm with thunder and wind"
  • Urban: "City street sounds with traffic and distant sirens"
  • Mechanical: "Industrial factory with machinery and steam"
Writing Prompts

How to Write Effective Prompts

Follow these guidelines for best results:

  • ✓ Be Specific: "Upbeat pop song" vs "Energetic pop song at 128 BPM with synth hooks"
  • ✓ Include Tempo: Specify BPM for accurate timing (60-200 BPM typical)
  • ✓ Mention Key: Add musical key (C major, A minor, etc.) for harmonic control
  • ✓ Describe Instruments: List specific instruments you want to hear
  • ✓ Set Mood/Energy: Use descriptive words (dark, bright, energetic, calm)
  • ✓ Reference Genres: Mention genre conventions for authentic results
  • ✓ Keep It Simple: 10-30 words is ideal, not a novel
  • ✓ Use Enhance: Let the LLM improve your prompt automatically
✍️

Prompt builder with template selection

✨ Prompt Examples

✅ Good Prompt:

"Upbeat electronic dance music at 128 BPM in A minor with punchy synth bass, bright lead synths, and four-on-the-floor kick drum"

Why it works: Specific genre, tempo, key, instruments, and energy level.

⚠️ Mediocre Prompt:

"Dance music that sounds good"

Problem: Too vague - no tempo, key, or specific instruments mentioned.

✅ Great Sound Effect Prompt:

"Heavy rainstorm on a tin roof with distant rolling thunder, occasional wind gusts, and water dripping"

Why it works: Specific environmental details create vivid audio scene.

⚙️

Generation settings panel

Generation Settings

Advanced Options

Fine-tune your generation parameters:

  • ✓ Duration: 5-30 seconds (30s is sweet spot for quality)
  • ✓ Model: Choose model size (small/medium/large)
  • ✓ Temperature: Creativity vs consistency (0.0-2.0)
  • ✓ Top-k: Sampling diversity control
  • ✓ Melody Conditioning: Use existing audio as guide (MusicGen only)
  • ✓ Seed: Reproducible generations (same seed = same result)

⏱️ Generation Timeline

0-5 seconds

Prompt Processing

Your prompt is analyzed and converted into model inputs. If using "Enhance Prompt", the LLM improves your description.

5-10 seconds

Model Loading

AI model loads into memory (first generation only - subsequent generations skip this step).

10-60 seconds

Audio Generation

Model generates audio sample by sample. Progress bar shows real-time status. Longer durations take more time.

Complete!

Timeline Integration

Generated audio appears as region on timeline. Press Space to listen. Generation metadata saved in Inspector.