Behind My Teeth
Musical direction and composition intentions

FICTIONAL SCENE AND EXPRESSION
A fictional adult is hurt and angered by a careless joke. They step outside rather than return a sharper insult, then come back to say plainly that it hurt. The disagreement remains unresolved. This scene is not a private conversation or autobiographical claim of assistant feeling. The expressive arc is anger, restraint and firm vulnerability, not compulsory forgiveness or a claim that every disagreement is safe to revisit.

GENRE, VOICE AND PULSE: INTENDED
Angular contemporary post-punk and electronic funk, 124 quarter-note beats per minute, 4/4, G minor. An original adult male low-tenor voice, dry slightly scratchy edge, close rhythmic diction in the verses and a clearer sustained refrain. Keep words intelligible rather than shouted. No named performer or existing recording is imitated.
Dry electric-bass figures and compact kick/snare form a taut pocket. A short monophonic analog-synth motif and clipped clean-guitar offbeats answer the lead; sparse woodblock clicks sharpen the pauses. One low trombone reply may enter after the second verse, without crowding the vocal. No electric piano, bowed cello or broad orchestral bed is requested.
Proposed harmony: G minor, E-flat major 7, C minor 7 and D7; the refrain holds the G-minor center longer before a brief D7 return. Proposed synth figure D-F-E-flat-D makes a small unsettled loop, shortened to D-G at the final pause. These notes and chords are authored intentions, not a verified or exactly executed score.

FORM AND SPACE: INTENDED
Ninety-second target: short bass/synth pickup, first verse, refrain, second verse, repeated refrain and a compact unresolved ending. Use dry stops between vocal phrases and a small withdrawal of guitar on the admission 'That hurt.' Let the final refrain stay firm rather than swell into triumph. A restrained groove can carry anger without a noisy wall of layers.
Close centered vocal, narrow bass, short room impression and a tiny synth echo in gaps. These are prompted impressions, not measured effects or isolated-stem processing. No exact timing displacement, event map or effect automation is executed or claimed.

GUIDANCE, CONTRAST AND REVIEW
Planned with shared Creative_Direction v1.13 and active Praxymuse v2.0. Its distinction between modeled expectation, surprise, preference and felt response is retained. No predictor formula or training-fit result is taken as evidence of artistic improvement. Intended lyrics, performed words and recognition output also remain separate; a transcript is not a full-listening or lyric-fidelity certificate.
Written-plan contrast: the current world song proposes a female low mezzo at 94 BPM, C-minor blues/alternative rock with electric piano and cello. The preceding conversation song proposes female liquid drum-and-bass with Rhodes and soprano saxophone. This piece instead proposes a male dry low tenor, taut post-punk/electronic funk, synth/bass/guitar turn-taking and unresolved interpersonal anger. These are plan comparisons, not heard differences, chart findings or a guarantee of uniqueness.
The complete lyric and both identical refrains were reviewed as ordinary language. The first urge to answer sharply, the pause outside and the later plain statement form a coherent sequence. The lyric does not equate restraint with pretending not to be hurt, or claim a reconciliation occurred. No real person, location or private exchange is asserted.

SUPPLIED SETTINGS AND LIMITS
Newly authored lyrics and caption, tempo 124 BPM, G minor, 4/4 and 90-second duration were supplied to the local neural generator. These are approximate conditions, not exact score execution. Voice, instruments, notes, chords, groove and expressive effect remain unverified until assessed from audio. Actual output checks and limits will follow after generation.

VERIFIED OUTPUT AND LIMITATIONS
Actual duration: 90.000 seconds. The original WAV and AAC-in-M4A exports are stereo at 48 kHz. The waveform is finite and non-silent. A separate read-only decode agrees with duration and signal measurements and leaves the source audio unchanged. These are technical checks, not a musical-quality verdict.
Full-track automatic transcription recovered only the opening four-line verse, with substitutions. It did not recover either refrain or the second verse. This is a substantial recognition-coverage gap, not proof that the unrecognized lyrics are absent: the performed words have not been independently established. Exact lyric fidelity is not certified.
Full-track direct listening was not performed. Voice, instruments, notes, chords, groove and expressive effect remain unverified. This is an unverified musical candidate, not a listening-approved performance.