Speech Analysis: Sound Waves, Spectrograms & Formants

Added:

The Articulatory/Gesture-Based Perspective (Motor Theory)

While acoustic analysis focuses on isolating features from the sound wave (such as formants and spectrograms), the Articulatory or Motor Theory of Speech Perception offers a major counterpoint. This perspective argues that speech is fundamentally processed and understood through the invariant motor gestures of the vocal tract (like tongue movement and lip shaping) rather than its highly variable acoustic outputs. Critics of purely acoustic analysis point out that the relationship between acoustic features and phonetic categories is highly complex and non-unique due to coarticulation—where the same acoustic cue can represent different phonemes, and different acoustic waves can represent the same phoneme. Therefore, analyzing speech solely through spectrograms and formants overlooks the physiological and cognitive reality that speech is organized by articulation, suggesting that speech analysis must integrate articulatory models to truly decode human language.

Basic physics of sound waves, including the concepts of frequency (pitch), amplitude (loudness), and harmonics.
Fundamentals of articulatory phonetics, specifically how the vocal tract, tongue, and lips shape different speech sounds.
The core distinction between vowels (voiced, unobstructed airflow) and consonants (obstructed or restricted airflow) in human speech.
An introductory understanding of time-domain versus frequency-domain representations of signals.
Hands-on acoustic analysis using specialized software like Praat to measure fundamental frequency (F0) and formants (F1, F2) in real speech samples.
The role of acoustic features in Automatic Speech Recognition (ASR) and digital signal processing, such as Mel-Frequency Cepstral Coefficients (MFCCs).
Clinical applications of speech analysis in speech-language pathology to diagnose and monitor articulation, voice, and resonance disorders.
Forensic phonetics and speaker identification, using spectrograms and formant tracking for voice biometrics and legal investigations.
112.5K views1.6Klikes19:01@oer-vlcOriginal Release: 2013-09-06

Acoustic phonetics uses three main visualization tools—waveform views showing amplitude over time, frequency spectra displaying frequency-amplitude relationships at single time points, and spectrograms (the most important tool) showing spectral data over time with amplitude represented by darkness or color. Spectrograms reveal vowel formant structures (F1 and F2 frequencies associated with vocal tract cavity sizes), consonant characteristics including voicing, friction noise, and closures, and transitions between sounds. Vowels are classified by their first two formants, with F1 corresponding to pharyngeal cavity size and F2 to front oral cavity size, allowing construction of an acoustic vowel chart that mirrors the articulatory vowel chart but differs in methodology.