Per-recording acoustic environment of 231 voice recordings made before and during a spaceflight: leading-silence noise floor, voice level and signal-to-noise ratio

One row per recording for three crew members, 36 pre-flight and 195 in flight, across sustained vowels and voluntary coughs. Columns: crew identifier, date, time, phase, task, ambient level, voice level, signal-to-noise ratio, leading silence duration, and the noise floor in dBFS measured from that silence. The protocol included an ambient calibration step whose decibel readings were not retained in the exported data; the delivered archive was searched for them without result. The noise floor was therefore measured from each recording's own leading silence, median duration 0.72 s. The signal-to-noise ratio is a difference between two levels measured through the same microphone in the same file and carries no calibration assumption. The absolute dB(A) columns are anchored to a published measurement of cabin ambient level and should be treated as indicative. Median signal-to-noise ratio falls from 28.55 dB in the pre-flight quarantine room to 16.70 dB in orbit for sustained vowels. That step alone costs 7.30 dB of harmonics-to-noise ratio, which accounts for three quarters of the observed pre-flight to in-flight change in cepstral peak prominence in the same crew. No audio is included. The recordings are not shared, for crew privacy and under the terms of the project agreement.