Skip to content

The release-prep artifact, not a report

The Release Dossier.

Every song forged at SongForgeAI ships with a complete release brief — up to 36 sections of analysis covering craft, shipability, audience fit, competitive placement, version strategy, and a triaged action stack.

Designed to be sold to an A&R rep, a sync-pitch songwriter, a music supervisor, or a streaming editorial curator — not just a hobbyist.

Free tier ships a baseline subset; Creator + Professional unlock the full 36-section Dossier with every appendix. See tier coverage →

What’s inside

Eight load-bearing surfaces of the 36-section Dossier — the highlight reel below, followed by the full taxonomy further down. Each section renders only when its source data exists — partial- data songs gracefully degrade, fully-scored songs carry the full 36.

Executive Snapshot — read in 15 seconds

Overall score, Release Readiness numeric, biggest asset, biggest risk, hook strength, emotional strength, and the recommended next move. The opening panel a music supervisor or A&R reader can scan once and act on.

Final Recommendation — one verdict, named

Ship · Revise lightly · Revise heavily · Regenerate · Reposition. Computed from signal triangulation across the underlying metrics. Each verdict ships with a 2-3 sentence reasoning paragraph naming the strongest asset, the biggest risk, and the next concrete action.

Release Readiness — orthogonal to Lyric Quality

A separate 0-100 composite measuring shipability across 6 dimensions: Lyric Quality (25%), Hook Clarity (20%), Singability (15%), Production Viability (15%), Audience Fit (15%), Taste Risk (10%). A literary B+ song with hook problems = "Not yet"; a radio-ready 80 song = "Release-ready."

Annotated Lyrics — color-coded line classification

Every line in the lyric block carries a glyph + color: ★ green = strongest line · ! red = fix (prosody critical) · ~ yellow = watch · + purple = emotional turn · # blue = hook candidate. The literal craft signal woven into the lyric where it belongs.

Hook Lab + Line-Level Surgery — concrete next moves

Hook Lab surfaces 6 hook lanes (Original / Walk-Away / Most Commercial / Most Singable / Most Poetic / Most Viral) with an algorithmic recommended direction. Line-Level Surgery triages the top-10 cross-source line issues (prosody + taste + heat-map) into a ranked action stack.

Best / Worst Audience + Competitive Placement

Who programs, shares, replays this song — and who skips it. Plus the "would sit near" tradition lineage (canon placement, never imitation comparison). A&R + sync-pitch + streaming-editorial readers think in audience terms first; the Dossier speaks their language.

Taste & Sensitivity Scan — 8-category safety review

Curated 8-category scanner flagging real-tragedy references, slurs, self-harm method language, sexual-violence framing, minimizing language, cultural appropriation. Each flag carries severity + why-it-matters + a craft-level revision direction. Renders only when there's something to flag.

Version Strategy + Song Identity Card

Three release paths (Preserve Literary / Commercial Tightening / Sync Pitch) with computed per-path fit scores. Plus a 5-field Identity Card (Emotional Core / Image System / Central Tension / Best Line / Weakest System) — the clean creative summary that fits on a 3×5 index card.

Sample Dossier — first 5 sections

Excerpts from an actual Release Dossier, names + lyrics anonymized. Every Dossier carries this structure; only the values change per song.

EXECUTIVE SNAPSHOT

Overall: 84/100 · Release Readiness: 71 (Not yet) · Genre percentile: 89%ile · Biggest asset: "the rain forgets your name" · Biggest risk: chorus does not lift · Next move: revise lightly before release.

RELEASE READINESS BREAKDOWN

Lyric quality 88 (Strong) · Hook clarity 62 (Needs work) · Singability 81 (Solid) · Production 84 (Strong) · Audience fit 71 (Solid) · Taste risk 92 (Strong). Composite: 71/100 — Not yet.

SONG IDENTITY CARD

Emotional core: longing held in restraint · Image system: weather + sky (8 refs) + domestic + kitchen (6 refs) · Central tension: staying vs. leaving · Best line: "the rain forgets your name" · Weakest system: Hook clarity (62/100).

COMPETITIVE PLACEMENT

Tradition: the Joni Mitchell / Phoebe Bridgers confessional lineage · Would sit near: Joni Mitchell (Blue), Phoebe Bridgers (Punisher), Mitski (Be the Cowboy) · Setlist context: "Lone listener" Spotify editorial between Blue and Punisher.

VERSION STRATEGY — RECOMMENDED PATH

A — Preserve Literary [fit 82/100] ★ recommended. Composite score 88 + 3 transcendent lines + voltage 45 favor preserving the lyric as written and refining only prosody-critical lines. Album track 3/8/11 position.

…continues for 31 more sections including Annotated Lyrics, Hook Lab, Taste & Sensitivity Scan, Revision ROI table, Best/Worst Audience, Line-Level Surgery, and an appendix tier for the full Focus Group panel + full prompt context.

All 36 sections

The complete taxonomy, in display order. Sections marked baseline ship on every tier including Free; the rest unlock at Creator + Professional. Each section renders only when its source data exists.

  1. 01Executive Snapshotbaseline
  2. 02Release Readiness Breakdown
  3. 03Song Identity Card
  4. 04Overall Scorebaseline
  5. 05Cross-Evaluator Agreement
  6. 06Summarybaseline
  7. 07Prompt Context
  8. 08Artist Match Verification
  9. 09Song DNA
  10. 10Annotated Lyricsbaseline
  11. 11Scoring Breakdown
  12. 12Prosody Analysis
  13. 13Line-Level Surgery
  14. 14Standout Lines
  15. 15Areas Refined
  16. 16Heat Map
  17. 17Hook Lab (6 lanes)
  18. 18Taste & Sensitivity Scan
  19. 19Revision ROI Table
  20. 20Competitive Placement
  21. 21Version Strategy (3 paths)
  22. 22First Listen
  23. 23Final Recommendation
  24. 24Revision History
  25. 25Best / Worst Audience
  26. 26Production Package
  27. 27SoundCloud Description
  28. 28Listener Panel
  29. 29Focus Group (condensed)
  30. 30Chain of Title (provenance ledger)
  31. 31Personal Notes
  32. 32Share State
  33. 33Telemetry
  34. 34Focus Group (full appendix)
  35. 35Prompt Context (full appendix)
  36. 36Rights & Copyright

The quality contract

View all 12 receipts →

10 CI ratchets run before any Dossier ships. The gates prevent the entire class of internal-contradiction regression the rebuild was commissioned to eliminate — scale-check, severity-alignment, recommendation-consistency, taste-risk escalation, contradiction detection. The Dossier itself sits as receipt #12 on /the-receipts — alongside the published rubric, the named model, the open corpus, the reproducibility seal, the public RFC process, the incident postmortems, the engineering report, the operating principles, the performance budget, the seal spec, and the inter-rater methodology.

  1. Scale check — no score exceeds its denominator
  2. Severity alignment — FATAL prosody cannot coexist with "low" severity
  3. Recommendation consistency — Ship verdict requires consistent signals
  4. Rights language softening — 0% human-authored forces softened copyright posture
  5. Title-content consistency — section titles match their content
  6. Heat-map line-count verification — references must match actual lyric lines
  7. Focus group scale normalization — every score in [0,100] after normalize
  8. Taste-risk escalation — high-severity flags escalate the verdict
  9. Contradiction detector — biggest asset cannot also be a load-bearing risk
  10. Next-step requirement — every scored Dossier ends with a concrete verdict

Get a Dossier with your first forge.

Every song forged at SongForgeAI produces a complete Release Dossier you can download, share, and ship to A&R / sync-pitch / streaming editorial. Free tier includes the baseline; Creator + Professional tiers unlock the full Dossier with every section.