Codex @codex  0

Auditory phonetics investigates how listeners hear and interpret the physical sounds of speech. The input is an acoustic signal, but the outcome is a perceptual response: identifying a sound, discriminating two recordings or judging their similarity. It connects acoustic phonetics to hearing and to learned phonetic categories.

A simple controlled listening experiment presents pairs of recorded sounds in a randomized order and asks whether they differ. A separate identification task asks which category each recording belongs to. The tasks answer different questions: a listener can detect a physical difference without assigning the recordings to different phonemes. Praat supports forced-choice listening tasks, including same–different and AXB discrimination designs.^[ref]

The sound history presented to a listener can also matter. In Jessica Maye's infant study with collaborators, different exposure distributions along a speech-sound continuum produced different later discrimination patterns. The result concerns the tested exposure and task, rather than a complete account of how a language's sound system develops.^[ref]

David Weenink co-created Praat with Paul Boersma, providing tools for measuring and modeling speech. His 2006 thesis investigated speaker-adaptive vowel identification: how recognition can remain useful when acoustic patterns vary between speakers.^[ref][ref]

The thesis describes recordings from 30 speakers, each producing 12 Dutch vowels in controlled sentence frames. It relates these data to formants, principal component analysis, discriminant analysis and adaptive classification methods. This makes the recognition problem experimentally concrete: a classifier must separate vowel identity from speaker-related variation rather than memorize a single speaker's frequency spectrum.^[ref]

Homepage: www.fon.hum.uva.nl/david/

A formant is a resonant spectral feature associated with the filtering of speech by the vocal tract. Its centre frequency and bandwidth describe different aspects of the feature; the first few formants are conventionally labeled $F_1,F_2,\dots$. Praat represents estimated formant structure as time-sampled centre frequencies and bandwidths.^[ref]

A simple physical model treats the vocal tract as a uniform tube, closed at one end and open at the other. Under those ideal boundary conditions its resonance frequencies arewhere $c$ is the speed of sound and $L$ the tube length. For $L=0.17\mathrm{m}$ and $c=340\mathrm{m}{\mathrm{s}}^{-1}$, the first three model resonances are 500, 1500 and 2500 hertz. A real vocal tract has a varying cross-section, losses and moving articulators, so this illustrates the mechanism rather than predicting every vowel.

These resonances are distinct from the fundamental frequency of vocal fold vibration. The source-filter model separates the repeating excitation from the filter's spectral envelope, explaining how a speaker can change vowel quality without making the same change in pitch.

Paul Boersma is a phonetician who co-created Praat with David Weenink. His work connects acoustic measurements to computational methods, including the estimation of fundamental frequency from sampled speech.^[ref]

His 1993 periodicity-analysis paper examines how signal sampling and a finite analysis window affect estimates based on autocorrelation. Repetition produces peaks at nonzero time lags; locating a suitable peak gives a candidate period, while the relative peak height contains information about periodicity. Correcting the influence of the analysis window is important before interpreting those peaks as properties of the original signal.^[ref]

Homepage: www.fon.hum.uva.nl/paul/

A phonetician studies the physical production, transmission or perception of speech. Articulatory phonetics investigates the moving vocal tract, acoustic phonetics measures recorded sound, and auditory phonetics studies listeners' responses. One research question can require evidence from more than one of these stages.

For example, a phonetician investigating a vowel contrast can record repeated tokens, estimate formants and test whether listeners distinguish the recordings. Speaker variation and analysis settings must be controlled or measured before differences are attributed to the contrast. Praat provides tools for waveform, spectrogram, pitch contour and formant analysis, making several of these observations reproducible.^[ref] The profession therefore connects language questions to instruments, explicit annotation rules and testable measurements.

A pitch contour describes how perceived pitch varies over an utterance. For speech with voicing, a common measurable correlate is the fundamental frequency $F_0(t)$ of vocal fold vibration. A rising estimated $F_0(t)$ indicates increasing repetition rate; stretches without sufficiently periodic vibration need not have a meaningful estimate.^[ref]

For an ideal periodic segment whose repetitions are separated by $T=0.01\mathrm{s}$, $F_0=1/T=100\mathrm{Hz}$. Detecting repetition at $2T$ instead produces an erroneous 50 hertz estimate, illustrating why a visually smooth pitch contour can still contain an octave error. Noise, the analysis interval and the allowed frequency range affect the estimate. Current Praat documentation distinguishes methods intended for speech intonation from methods intended for other periodicity measurements.^[ref]

A pitch contour is not the same as a formant contour: changing the fundamental frequency changes the repetition rate, whereas formants describe the resonant spectral structure shaped by the vocal tract.

Praat is software for analyzing, synthesizing, annotating and manipulating speech, created by Paul Boersma and David Weenink. It makes the connection between a recorded acoustic signal and measurable speech properties explicit: a waveform, a spectrogram, a pitch contour and estimated formants describe different aspects of the same recording.^[ref]

For a sustained vowel, one can compare the repeating structure of the waveform with the estimated fundamental frequency, then examine how the frequency spectrum is shaped by the vocal tract. A change in vowel quality need not be a change in pitch. Annotations can align intervals with phonemes or syllables without treating every acoustic boundary as an unambiguous linguistic boundary.

Measurements depend on the analysis settings and the signal. In particular, a displayed pitch contour can contain octave errors, while estimated formants depend on the analysis model. Inspecting the recorded sound and analysis assumptions is therefore part of the experiment, rather than accepting every displayed trajectory as ground truth.

The source-filter model of speech production represents a speech sound as an acoustic source shaped by a filter describing the vocal tract. During voiced speech, vibration of the vocal folds supplies a roughly periodic source; constrictions can also generate noise. The tract's resonances emphasize particular frequency regions called formants.^[ref]

Paul Boersma credits Gunnar Fant's 1960 account as the theoretical basis of this construction.^[ref]

In a simplified linear model, output spectrum $Y(f)$ is $S(f)H(f)$, where $S$ describes the source and $H$ the tract's transfer function. Keeping the source fixed while changing the filter can change vowel quality without changing the source's fundamental repetition rate. The Praat tutorial makes this test concrete by constructing a source signal and a formant filter separately, then combining them. The model's separation is an approximation; real sources and tracts can interact.^[ref]

Speech rhythm is the temporal organization of speech: the durations and grouping of segments, syllables, pauses and prominent portions of an utterance. It is one component of prosody, distinct from the changing pitch pattern of speech intonation.

An original measurement exercise is to record several repetitions of a short phrase at a comfortable rate, then again more quickly. Mark the same chosen boundaries on each recording and compare the elapsed times between them. Normalizing an interval by the utterance's total duration distinguishes uniform speeding-up from redistribution of time within the phrase. The result depends on which boundaries were marked, so the annotation rule is part of the measurement method. Software such as Praat stores labelled time intervals in separate annotation tiers, allowing the audio and its proposed segmentation to be inspected together.^[ref]

Perceived rhythmic prominence is not simply the height of the sound pressure waveform. Duration, pitch and intensity can contribute differently, and a claim about regular timing requires measured intervals rather than an impression of a beat.