FRAME | After the Lights Go Out

FRAME | After the Lights Go Out — English night-gallery guide, 38.64 s, a real MP3 (MPEG-2 Layer III / v2, 128 kbps, 24 kHz mono, ID3-tagged).

Voice: TTS-1, "fable" at speed 0.85 — unhurried, close and curious, as asked. Your copy is read as supplied, with no music bed, no added lines and no real collections named.

Measured on the delivered file itself:

• Duration: 38.640 s — inside 35–45 s
• Start: first word at ~0.09 s, no dead lead-in
• End: last word finishes at 37.94 s, then a 0.64 s clean tail — nothing clipped
• Volume: −19.1 LUFS integrated, −1.9 dBFS sample peak / −1.7 dBTP true peak, mean −21.4 dB, LRA 9.5 LU — a healthy, unclipped narration level, lifted from the raw take's −26.0 LUFS

What I changed, and why: the raw take ran 29.2 s, under your minimum. Rather than speeding the words back up, I kept the voice as recorded and gave each of the five sentence breaks a ~1.75 s breath of silence, so the guide lands at 38.6 s and reads as deliberately unhurried — visitors get time to look.

One honest gap: the attachment-analysis service refused its credentials during this run, so I could not machine-transcribe the audio to re-check every word or judge pronunciation and timbre by listening. Duration, edges and level above are measured from the file; wording fidelity rests on the text I sent to the voice, not on a second transcription.