The mel scale appears discredited but its underlying effects still hold
Evidence
Criticisms of the mel scale are many, ranging from bias in experimental techniques to wide variability within and across listeners with normal hearing to the fact that it does not correspond to the musical scale and its associated perceptions. In fact, if we conceptually consider the implications of applying subjective scaling procedures to a perception like pitch, further questions arise. Unlike loudness, pitch does not have a “magnitude,” and the mel scale ignores pitch strength (the degree to which a pitch is clear and strong) and pitch chroma (the perception of similarities of octaves), which are also important aspects of the pitch percept. The data relating the mel to frequency are suspect and potentially problematic. As a consequence, the mel is rarely used in the psychophysical literature, although it is still used in the field of speech perception and by acoustic engineers. However, the experimental result that increasing frequency leads to increasing pitch is important and has been replicated numerous times. The difference in the shape of this function at low vs. high frequencies is also repeatable, and suggests different physiological mechanisms responsible for the perception of pitch of sounds of low versus high frequencies. — Jennifer Lentz, Psychoacoustics- Perception of Normal…, 1245
Categorization
Category: Contrarian or Surprising Claims
- XREF: The finding that low vs high frequency pitch may rely on different physiological mechanisms ties to place vs temporal/rate coding theories of pitch that likely recur in later chapters.
Related
- Pitch has no magnitude like loudness, so magnitude scaling misapplies to it. (similarity 0.95)
- The mel scale was derived by asking listeners to double perceived pitch (similarity 0.87)
- Musical pitch perception spans roughly 30 Hz to 4 kHz. (similarity 0.87)
- Pitch lacks a magnitude dimension, complicating its subjective measurement (similarity 0.85)
- The mel scale defines pitch by assigning 1000 mels to a 1000-Hz tone (similarity 0.85)
- Pitch is perceptual while frequency is acoustic, and not all sounds have pitch (similarity 0.85)
- The frequency-pitch mapping was established before models of pitch existed (similarity 0.85)
- Pitch depends only on spectrum and periodicity, not volume or direction. (similarity 0.83)
- Pitch is coded by different mechanisms above and below roughly 3 kHz (similarity 0.83)
- Perceived pitch is governed mainly by fundamental frequency, not loudness or duration (similarity 0.83)
- Raising intensity lowers pitch at low frequencies but raises it at high frequencies. (similarity 0.83)
- Sounds missing many harmonics still evoke a weak pitch sensation (similarity 0.83)
- Even nonperiodic noise sounds convey a weak pitch sensation based on spectral energy distribution (similarity 0.82)
- Human pitch sense spans 0 to roughly 5000 Hz, distinct from partial perception (similarity 0.82)
- Loudness slightly but measurably alters perceived pitch of tones (similarity 0.82)