Speech token recognition was tested across six signal-to-noise ratios from 12 to -15 dB

Evidence

Six recordings of each CV (three spoken by a male, three spoken by a female talker) were taken from the Danish nonsense syllable speech material collected by Christiansen and Henrichsen ( 2011 ). Six SNR conditions (12, 6, 0, ‑6, ‑12, and ‑15 dB) were created by fixing the noise sound pressure level (SPL) to 60 dB and adjusting the speech SPL. One particular white masking noise waveform with a duration of 1 s was generated for each speech token in each SNR condition and mixed with it such that the speech token onset was temporally positioned 400 ms after the noise onset. — Dijk, Pim van, Physiology, Psychoacoustics and…, 5580