Prompt file imported from asume21/Codedswitchmonatize (
.windsurf/workflows/ai-song-production.md). Copyright stays with the author.
Overview
This workflow guides a user through the full song-making lifecycle using the app’s AI features: ideation, lyrics, structure, patterns, audio generation, vocal integration, mixing, and mastering.
Steps
-
Set project context
- Decide genre, mood, BPM, key (or leave to AI).
- Create a new session; ensure env keys (Suno/Replicate, OpenAI/Grok, ElevenLabs if used) are set.
-
Concept & lyrics
- Use Lyrics Generate (
/api/lyrics/generate) with theme/genre/mood. - Refine with Lyrics Analyze (
/api/lyrics/analyze) to improve rhyme, syllable balance, and theme fit.
- Use Lyrics Generate (
-
Song plan / structure
- Use AI Song Plan (
/api/ai/song/plan) to get sections (intro/verse/chorus/bridge) and durations. - Pick target BPM/key from the plan (or set manually).
- Use AI Song Plan (
-
Pattern creation (editable MIDI)
- Use Astutely pattern to draft drums/bass/chords/melody.
- Review in the Piano Roll; edit notes if desired.
-
Real audio generation
- Trigger AI audio (Suno/MusicGen) for the arrangement.
- If using “Edit in Piano Roll” after Suno: run Extract Patterns (
/api/songs/extract-patterns) and load both editable notes and the reference audio track.
-
Vocal path (optional)
- Generate vocals with ElevenLabs (voice ID) or in-app vocal tools.
- Align vocal stem to BPM/key; import as audio track.
-
Stems / layering (optional)
- Run stem separation if you need vocal/instrumental splits for further mixing.
-
Mix & balance
- In the studio, set levels, pans, basic EQ/comp as needed.
- Use transport loop over the main section to balance.
-
Master
- Export master via the app’s export flow; verify loudness and headroom.
- Spot-check beginning/ending trims and fades.
-
Finalize
- Save session, render final WAV/MP3, and archive stems.
- (If deploying) push changes/metadata to repo or asset store.
Notes
- If local Llama (Ollama) is available, Astutely will prefer it; otherwise it falls back to cloud.
- Keep BPM/key consistent between patterns, audio generation, and vocals for best alignment.
- For best quality: run a final listening pass on headphones and speakers.