Skip to Content

Department of Linguistics

SPEECH ACOUSTICS

Frequency Transformations of Speech
Pitch Perception and
Frequency Discrimination

Robert Mannell

Pitch Perception

In speech perception, pitch is the perceptual correlate of fundamental frequency (F0). Pitch perception is the relative perception of the fundamental frequency of two separate sounds, such as two pure tones played one after the other (or simultaneously in separate ears). We build up a picture of the perceptual scale upon which pitch perception is based by playing experimental subjects a reference tone and asking them to adjust a second tone until it sounds twice as high or half as high in pitch. By varying the reference tone and by testing a large number of subjects we can build up a scale which relates physical F0 (in Hertz) to a subjective perceptual scale of pitch. The unit of pitch is called the "mel". 1000 Hz was arbitrarily selected and its pitch in mels was chosen to be 1000 mels. 500 mels is perceived as half the pitch of 1000 mels, whilst 2000 mels is perceived as being twice the pitch of 1000 mels, etc.

It is clear from this graph that the relationship between frequency (in Hertz) and pitch (in mels) is not linear. This curve describes the relationship between Hertz and mels for an average normal hearer.

This scale does not relate to our perception of the frequency components of complex spectra, such as speech spectra, but rather to the perception of the fundamental frequency. Some studies of intonation and tone use the mel scale rather than the Hertz scale as it provides a better representation of the perceptual significance of the difference between two F0 contours.

You should note that the mel scale is not the same as the musical scale in western music (or for that matter, other musical traditions). The western musical scale is based on simple ratios between frequencies and is related to the idea of harmony (or "consonance"), where certain ratios of F0 are selected so that some harmonics of two (or more) sounds occur at identical frequencies when played simultaneously or in sequence.

For more information on pitch perception, click here.

Frequency Discrimination

Frequency discrimination usually refers to the determination of just noticeable differences in frequency. Very often such measurements refer to just noticeable differences in pitch (ie. of fundamental frequency). For example, some studies suggest that we should be able to detect a difference in F0 of 2 Hz at 1000 Hz, but not smaller differences. Some studies have examined just noticeable differences in formant frequency and found that changes of about 3-5% in formant frequency can just be detected.

For more information on frequency discrimination, click here.