In these sources they are talking about the fundamental frequency of human Speech F~0~. Which is one important indicator but I don’t think this band is sufficient for comprehension. Record yourself, apply a low-pass filter with cut frequently at 300Hz and check for yourself if this is enough to understand. It might be enough to make out some words, but I doubt it’s enough to hear clearly.
anomnom@sh.itjust.works 7 hours ago
Maybe machine learning can infer the missing wave information (like Fourier transforming partial waveforms??