Voice & POV Integrity
Narrator consistency, perspective clarity, and a credible speaker. Does this sound like one person talking?
What good looks like
A distinct human presence. Word choices, diction, and references that belong to one coherent narrator.
How SongForgeAI scores it
narrator-profile.ts (Build 951) extracts gender / age / relational tuple from the lyrics and flags cross-section contradictions ("my wife + my husband", "I'm 16 + I'm 40").
Sub-criteria
Inside VoiceNamed sub-concepts the eval engine considers when computing this metric’s score. Each one is a discrete signal — ignoring it pulls the metric out of band even when the other dimensions look intact.
POV Consistency
A narrator stays internally coherent across sections. The speaker who said "I take the long way home" in V1 cannot be the speaker who says "we built this city together" in V3 unless the song earned the shift. POV consistency tracks gender / age / relational tuple / diction register; intentional POV switches (gospel call-and-response, K-pop multi-voice, hip-hop features) count as consistent when each voice is internally stable and switches mark structural moments. v1.2.0 of the rubric formalized "intentional POV" as the metric's lens; this sub-criterion names the underlying signal.
Signalsnarrator-profile.ts (B951) extracts gender / age / relational signals per section; cross-section contradictions flagged ("my wife" + "my husband", "I'm 16" + "I'm 40"). Diction-shift detection catches register drift (a mechanic doesn't suddenly quote Rilke). Intentional POV switches are recognized via section-marker patterns + collaborative-form signals.
Failure looks likeA speaker who's 24 in V1 ("I'm too young to know what I want") and 50 in the bridge ("I've lived enough lives to know"); or a narrator who speaks in workplace argot for two verses and suddenly invokes Stoic philosophy in the chorus without earning the shift; or a song that switches from first-person singular to first-person plural in the bridge for no structural reason.
Example score
81/100on this metricNarrator is consistent across V1 ("I take the long way home") and V3 ("I'm fine. I'm fine.")
WhySame speaker, same emotional register, same diction throughout. Would score higher if the bridge introduced a credible POV expansion (e.g., addressing a specific person by name) without breaking the established voice.
One representative example. Real scores carry a reproducibility seal — verify at /scoring/standard.
Common failure mode
POV drift — the speaker sounds like three different people across three verses.
What it looks like in the wild
- Verse 1 and verse 3 clearly spoken by the same person in the same emotional weather.
- Diction that matches the character — a mechanic doesn't quote Rilke.
- The narrator credibly knows what they claim to know.
Why it’s in the Expression tier
Does it say something worth hearing? Specificity, originality, truth, and voice.
The Expression tier contributes 40% to the final composite score. The tier weight is distributed across its member metrics — no single metric dominates the composite.
Other Expression metrics
Lyrical Specificity
#5Concrete imagery, sensory detail, proper nouns, time anchors. The opposite of abstract generalities.
Imagery Originality
#6Fresh metaphors, defamiliarized objects, governing images that haven't been written to death.
Emotional Truth
#7The ring-test: does it feel true? Earned emotion, unforced vulnerability, no borrowed sentiment.
The full rubric
Related reading
Narrative Voice: the metric AI lyric tools fail hardest
The founder-voice essay on the Voice metric — what makes a narrator recognizable, why AI fails this one hardest, how the gauntlet fixes a low Voice score.
Read the essayWriting Breakup Songs Without Cliché
Narrative Voice is what separates "a breakup song" from THIS narrator’s breakup song.
Read the guidePublishing the Lyric Scoring Standard v1.0
The rubric itself, open-sourced under CC BY 4.0.
Read the essayWhy the default score is 50, not 75
Why the Gravity Rule keeps every metric honest.
Read the essayAnatomy of a Forge: one song from prompt to final score
See this metric scored on a real forge run, alongside all 11 others.
Read the essaySee this metric scored against real songs
Every song forged through SongForgeAI is scored on Voice. Browse the leaderboard or forge your own to see how lines land on this axis.