Praat Tutorial: Measure Vowel Formants (F1 & F2) in Acoustic Analysis

Added:

Limitations of Static F1/F2 Tracking and the Whole-Spectrum Alternative

While measuring static F1 and F2 at a single time point (typically the vowel midpoint) is the traditional standard in acoustic phonetics, this approach has significant limitations. First, LPC-based formant tracking in Praat is notoriously prone to errors in high-pitched voices (such as children's or high female voices), nasalized vowels, and creaky voice, where the algorithm often merges or misidentifies formants. Second, static measurements ignore 'Vowel Inherent Spectral Change' (VISC)—the dynamic trajectory of a vowel over time—which perceptual studies show is crucial for human vowel identification. Critics and modern phoneticians advocate for alternative methods, such as whole-spectrum analysis (using Mel-Frequency Cepstral Coefficients or spectral moments) and dynamic trajectory tracking. These alternative frameworks argue that vowel identity is encoded in the continuous, global spectral shape of the acoustic signal rather than isolated, static formant frequencies.

Basic concepts of acoustic phonetics, particularly the source-filter theory of speech production.
An understanding of what formants are and how they relate to vocal tract resonance and vowel quality (e.g., how tongue height and advancement affect F1 and F2).
Familiarity with the Praat user interface, including how to open audio files and navigate the Sound editor window.
How to read a basic spectrogram, including the representation of time, frequency, and amplitude/intensity.
Scripting in Praat to automate formant extraction and analysis for large acoustic datasets.
Vowel normalization methods (such as Lobanov or Nearey) to control for anatomical differences among speakers.
Visualizing vowel spaces (plotting F1 vs. F2) using statistical software like R, Python, or specialized phonetic tools.
Analyzing dynamic formant trajectories (such as measuring diphthongs or coarticulation effects over time rather than at a single static point).
44.5K views486likes3:11@linguistiklaboralbert-ludw3514Original Release: 2019-05-04

In Praat software, vowel formants (F1 and F2) can be measured by identifying the gray shades in the spectrogram; F1 is inversely related to vowel height (high vowels have lower F1 values), while F2 is directly related to vowel frontness (front vowels have higher F2 values), and measurements are obtained by clicking on stable formant regions and selecting Format Listing to display their values.