Complete reference for AI MUSIC PROMPT ARCHITECT v9.7. For a downloadable PDF version, use the ⬇ HOW TO USE → Download PDF Manual button inside the app.
AI MUSIC PROMPT ARCHITECT requires a valid Anthropic Claude API key set as the environment variable ANTHROPIC_API_KEY before launch. The key must begin with sk-ant-.
Python 3.10+ and Flask required. Install dependencies: pip install flask anthropic reportlab
Run: python app_complete_autoport_v9_7.py — the app auto-detects a free port starting from 5000 and opens in your default browser.
5 tabs at the top: 🎵 Music Prompt, 🖼 Image Analysis, 🎭 Morph Your Song, 🔥 Album Forge, 🧬 Lyric Alchemy. Two status panels fixed at the bottom: Network & API Status (left) and Live Activity Log (right).
Generate a complete 9-section Suno prompt from scratch. Fields: Track Title (optional, UPPERCASE, max 50 chars), Artist Name (default "Obiriec"), Music Genre (be specific — "Noir Spy Orchestral" beats "Electronic"), Mood/Atmosphere (sensory language, 1–3 sentences).
Weirdness (0–100): Controls experimental direction. 0 = conventional genre structures. 100 = avant-garde, genre-bending. Sweet spot for professional music: 30–60. Values above 70 auto-trigger instrumental mode in SURPRISE ME!
Style Influence (0–100): How strongly genre archetypes, production signatures and iconic references influence the output. Both sliders are always active — they apply to SURPRISE ME! as well as GENERATE.
SURPRISE ME!: Ignores all text fields. The AI creates a complete random brief and builds the full 9-section output. Weirdness and Style Influence are the only user controls in a random generation.
Drag-drop or browse any image (JPG, PNG, WEBP, GIF, max 10MB). The AI analyzes colors, composition, mood, narrative content. Optional description field adds context not visually obvious.
Interpretation slider (0 = Literal, 100 = Creative): At 0, visual content directly translates to sound. At 100, the AI uses the image as an emotional metaphor. For album art: 0–30. For creative inspiration: 60–100.
Images are sent as base64 and never stored permanently. Expect 15–40 seconds for analysis — the payload is significantly larger than a standard prompt generation.
Paste your own original lyrics (max ~4500 chars). Select a Target Genre, Texture & Mood, Vocal Gender (🎲 for random), and Vocal Character. The optional Energy Directive is a free-text field for production energy — e.g. "Builds from whisper to explosion".
The output contains three independently copyable sections: Transformed Lyrics (rewritten with Suno section tags), Style Prompt, and Exclude Styles.
⚠ Only paste material you own or have rights to. Do not paste copyrighted commercial recordings.
Enter Artist, Album, Track, Era, Context — at least one required. EXPLORE runs a web-search-enhanced analysis and returns Sonic Profile, Production DNA, Style Tags, and Avoid Tags. Click USE AS ALBUM DNA to populate the Album Style Prompt automatically.
Number of Tracks: A 12-track album = 13 API calls (1 brief + 12 track prompts). Allow several minutes.
Album Cohesion (0–100%): 0% = independent sonic experiments. 100% = maximum stylistic unity.
Instrumental Ratio (0–100%): Proportion of instrumental tracks. Distribution is random each forge. 🎼 marks instrumental tracks in the tracklist.
Track navigation: ◀ PREV / NEXT ▶. Visited tracks are cached — instant re-navigation. RE-GENERATE THIS TRACK regenerates only the current track.
⚡ Magic Compilation: Fully random independent tracks with no thematic coherence. Cohesion setting ignored.
The most structured creative workflow. Each step feeds the next. Step results are cached — go back and change any step without losing unaffected downstream work.
Step 1 — Creative DNA: Intensity Level + Narrative Seed + Internal Conflict + Driving Force + Resistance + (optional) Resolution Arc. Be concrete and specific — this is the lyric-writing brief.
Step 2 — Narrative Paths: AI generates 3 directions (Outer/Inner/Symbolic). Edit or select. Selecting a different narrative invalidates steps 3–8.
Steps 3–5: Story Voice (1st/2nd/3rd/Omniscient) → Chorus Profile (Arena/Close/Loop/Shadow) → Voice Architecture (Lead Voice Lock + optional Backing Voice)
Step 6 — Genre Universe: 60 categories across 10 families. Changing genre invalidates steps 7–8.
Step 7 — Style Refinement: 4 AI-generated production sub-styles for your chosen genre. Cached per genre.
Step 8 — Output Draft: Full lyrics + Production Brief. Click ⚗️ BUILD FULL SUNO PROMPT to assemble the complete 9-section output.
🎵 Style Prompt → Suno Style field. Max 1000 chars. Suno silently truncates beyond this — the app enforces the limit automatically.
🚫 Exclude Styles → Suno Exclude field. Format: "EDM, Trap, Auto-Tune" — NOT "No EDM, No Trap". The "No" prefix reduces effectiveness in Suno.
📝 Lyrics / Structure → Suno Lyrics field. Max 4500 chars. Vocal tracks use [verse], [chorus], [bridge], [outro] tags. Instrumental tracks use [section name](description) format — e.g. [verse](sparse piano, 4/4, hesitant phrasing).
🖼 Album Cover Prompt: Ready for Midjourney, DALL-E, Stable Diffusion, Ideogram. Paste directly.
Deploy order in Suno: Style Prompt → Exclude Styles → Lyrics. Set title separately. Generate 2 variants — the second is often stronger.
Every generation passes through the Obiriec quality pipeline: ANTI_AI_INSTRUCTION injected in 5 prompts, sanitize_banned_terms() with 60+ rules applied to all output, character limits enforced in every prompt instruction.
Banned terms (never generated): neon, tapestry, odyssey, testament, gossamer, labyrinth, soundscape, pulsating, mesmerizing, hauntingly, timeless, transcendent, seamlessly blend, heart and soul, musical journey, unique blend, ethereal, and 45+ more.
Lyric quality — 8 criteria enforced: Sensory specificity · Conflict coherence · Syntactic variation · Chorus as thesis · Sound play · Breath management · Show don't tell · Irony subtlety.
Network & API Status (bottom-left): 4 checks every 30 seconds — Internet (TCP 8.8.8.8:53), Anthropic API (TCP api.anthropic.com:443), Web Search Tool (inferred), API Key (format check). Green = all good. Click ⟳ RUN TEST for immediate check.
Live Activity Log (bottom-right): Real-time SSE stream. Blue = API calls, Green = OK, Yellow = WARN/retry, Red = ERROR, Grey = INFO. Auto-reconnects if SSE connection drops. Click ✕ Clear to reset.
Style Prompt density: 15–25 specific non-redundant tags outperform 40 vague ones. Priority order: genre → tempo → instruments → production era → vocal style → mood qualifiers.
Lyric Alchemy for best lyrics: Consistently produces higher-quality results than Music Prompt because the 8-step structure forces disciplined creative process rather than free-text description.
Retry on 529 overload: System retries 3 times automatically (3s, 8s, 15s delays). Do not click again. Anthropic overload events typically resolve within 1–2 minutes.
Instrumental mandatory format: [section name](description) — e.g. [bridge](modulating piano, tension build, 8 bars, ascending line). This is required by Suno for instrumental tracks.