Direct answer
A healthy young human ear hears sound from about 20 Hz to 20,000 Hz (20 kHz), a span of roughly ten octaves [1]. Sensitivity is not flat across that span: hearing is sharpest between about 2 kHz and 5 kHz, which is where much of the energy in speech consonants sits [1]. Above 20 kHz you are in ultrasound territory, and below 20 Hz you are in infrasound — neither is normally audible, though very low frequencies can still be felt as vibration [1]. In practice, most adults top out well below 20 kHz, often nearer 15–17 kHz, because the upper limit drops steadily with age and noise exposure [3].
What Frequency Range Can the Human Ear Actually Hear?
The 20 Hz–20 kHz figure is the conventional textbook range, and it is worth understanding what it actually describes: the range over which a young, undamaged cochlea can register a tone at a detectable level, not the range at which every sound is equally loud [2]. At the extreme low end, a 20 Hz tone has to be delivered at considerable amplitude before it is perceived at all, and much of what people call "feeling the bass" at a concert is tactile and vestibular rather than purely auditory [1]. At the high end, the threshold rises steeply, so a 16 kHz tone that a teenager hears easily may be silent to their lecturer [2].
Where hearing matters most for daily function is a much narrower band. Speech intelligibility is carried mainly by energy between roughly 250 Hz and 8 kHz, which is precisely why clinical audiometry tests octave frequencies from 250 Hz through 8000 Hz rather than sweeping the full 20 kHz span [2]. A person can therefore have a technically "reduced" high-frequency range and still follow conversation perfectly in a quiet room, then struggle badly in a noisy seminar or a lecture hall with hard reflective walls [2].
It also helps to separate frequency from loudness, because the two are judged by different limits. Frequency range describes which pitches reach the brain; the threshold of hearing and the threshold of pain describe how much energy is needed to register them, and both thresholds shift with frequency [1]. That is why a low-frequency rumble can be inaudible yet physically uncomfortable, while a mid-frequency tone at the same measured level sounds clearly loud [1]. For a student reading a Turnitin AI report, the useful analogy is that a single headline percentage hides a lot of internal structure — the same way "20 Hz to 20 kHz" hides a very uneven sensitivity curve [2].
Does Human Hearing Range Change With Age and Volume?
Yes, and the change is predictable. Age-related hearing loss — presbycusis — is typically gradual, affects both ears, and begins at the highest frequencies first, so the practical ceiling of hearing slides downward decade by decade [3]. Because the lost frequencies are above the main speech band, the earliest symptoms are rarely "I cannot hear anything"; they are more often difficulty separating one voice from background noise, needing the volume raised, or asking people to repeat themselves in group settings [3]. This is why a 20-year-old and a 60-year-old can both pass a basic conversation test while having very different measured high-frequency limits [3].
Loudness matters because it is the main modifiable risk factor. Sustained exposure to loud noise accelerates the same high-frequency decline that ageing produces, and the damage accumulates without pain as a warning signal [3]. The practical rule most audiology clinics give is distance and duration: the further you are from the source and the shorter the exposure, the lower the risk, and ringing or muffled hearing afterwards is a sign the exposure was too much [3]. Headphone volume during long study sessions is the exposure pattern most students actually control.
There is also a measurement point that trips people up when they compare themselves to online hearing tests. A phone speaker, laptop driver, or earbud often cannot reproduce 18–20 kHz at a reliable level at all, so a "failed" high-frequency test may be measuring the hardware rather than the ear [2]. Clinical testing controls for this with calibrated transducers and background-noise limits, which is why an audiogram from a clinic and a web test are not interchangeable [2]. If you genuinely want to know your range, the number to trust is the one produced under calibrated conditions [3].
Can a Turnitin AI Detector Tell Human-Sounding Writing Apart From AI Text?
This is where the phrase "sound human" gets borrowed from acoustics into academic integrity, and the borrowing causes real confusion. Turnitin's AI writing detection does not listen for a voice or judge whether prose "feels" human; it analyses the statistical and structural patterns of qualifying text and reports the percentage of that text its model classifies as AI-generated [4]. Crucially, that percentage is a signal about text, not a finding about authorship — the report itself is designed to be read alongside context, drafts, and supervisory judgement rather than treated as proof [4].
Two features of the report explain most of the anxiety students report. First, only qualifying text is scored, so short submissions or heavily quoted documents may produce a low or absent percentage that says little about the writing itself [4]. Second, when the detector's confidence falls below its internal threshold, the score is displayed as *% rather than an exact figure — a low-confidence signal that should not be read as either a clean bill of health or an accusation [4]. A hand-written, deliberately plain essay can therefore land in the same ambiguous band as a lightly edited AI draft, because both are short on the long-range predictability the model keys on [4].
The practical takeaway is that "sounds human" is not a property the detector measures, so no amount of stylistic polish guarantees a particular number. What you can control is knowing your own draft's result before it becomes a submission, which is exactly what a pre-submission check is for: it shows you the same similarity and AI report structure your institution sees, so a surprising score becomes something you can investigate and revise rather than something you discover after the deadline [4]. Treat the score as a diagnostic reading of your text — the way an audiogram is a reading of your ear — and the ambiguity becomes manageable rather than alarming [4].
If you would rather see that reading before your tutor does, turnitin0 gives you the same report format your institution works from — AI score, flags, and similarity matches on your own draft, back in minutes and without adding your file to Turnitin's student paper database.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
What is the lowest sound a human can hear?
About 20 Hz for a young, healthy ear, and only at a fairly high amplitude — very low frequencies are often felt as vibration before they are heard [1]. Below that is infrasound, which is not normally audible [1].
Can humans hear above 20 kHz?
Not through the ear in any useful way; 20 kHz is the conventional upper limit, and most adults hear nothing above roughly 15–17 kHz [1][3]. Anything higher is ultrasound [1].
Why do older people lose high frequencies first?
Presbycusis affects the highest frequencies earliest, which is why the first complaints are usually difficulty following speech in background noise rather than total silence [3]. Noise exposure accelerates the same pattern [3].
Does a low Turnitin AI score prove the text was written by a human?
No. The report shows the share of qualifying text the model classifies as AI-generated, and low-confidence results appear as *% rather than an exact figure, so it is a signal to interpret with context, not proof of authorship [4].
Can a Turnitin AI report flag writing that I wrote entirely myself?
It can, particularly with short or heavily formulaic text, because the detector scores patterns rather than verifying authorship [4]. That is the main reason to preview your own report before submitting [4].