Why a low male voice survives 4× playback

📉① Clarity does not collapse at the same rate

Time-compress any recording and it gets harder to follow — but the slope differs. In time-compressed listening experiments, intelligibility for female speakers fell more steeply as the compression ratio rose, and the increase in reported listening difficulty was consistently smaller for male voices. The faster you go, the more a low male voice holds up by comparison.

📐② Vowel space — male speakers have more room

Under extreme acceleration the mouth cannot finish moving before the next syllable arrives, so formants crowd toward the centre. How far apart the vowels started out becomes the defence.

• Female · ~13.8–14.2 BarkA comparatively narrow vowel-space perimeter, so modest speed increases are enough for phoneme boundaries to overlap and vowels to be confused.
• Male · ~15.3–16.0 BarkA wider vowel space, so even when formants compress at 3–5× there is still physical distance between vowels — discrimination degrades far more slowly.

⚙️③ How it meets the time-stretch algorithm

Algorithms that change speed while holding pitch (WSOLA and relatives) cut the waveform into fragments and splice them back so the phase lines up. Female voices carry a higher fundamental (f₀) and denser harmonic spacing, so when the algorithm misses that fine periodicity the joins produce a shimmering metallic artefact (phasiness). A steady low male voice in the 95–105 Hz range keeps its acoustic skeleton through the same processing, and decodes more cleanly as a result.

🎯Not a preference — a design constraint

At 1× the speaker’s voice is close to irrelevant. From about 3× upward, a steady low male voice is a practical shield against the acoustic distortion of time compression and the noise the algorithm adds. That is why BrainHz keeps its speed-listening voice low and even — and why pairing it with lossless WAV is what makes the effect complete.