Autocorrelation, proposed in 1951, remains essential to pitch perception theory

Evidence

The idea that autocorrelation is what determines pitch came rather late, only in 1951, suggested by J.C.R. Licklider. The notion seems to have had a rather lukewarm reception in the literature ever since, yet autocorrelation is what pitch estimators use in many sound analysis programs, such as Praat and Audacity. Physiologically, it is not clear whether autocorrelation is literally computed in a neural circuit or merely strongly related to whatever is. The autocorrelation idea is an example of a temporal theory of pitch perception. We will expand on how autocorrelation may be used to determine pitch shortly, but it works so well in so many circumstances that it seems safe to say this: Beware of any theory of pitch perception that entirely leaves out autocorrelation. — Eric J. Heller, Why You Hear What You Hear- An…, 2073

Categorization

• XREF: Connects to the earlier autocorrelation definition in chapter 4 that this chapter’s argument builds on.
• QUESTION: Whether autocorrelation is literally computed in neural circuits or merely correlates with neural processing is unresolved and worth chasing.