šŸŽ™ļø Vowel Space Visualizer

Vowel Space and Formant Trajectory Visualization Tool

File Upload

Select the frequency scale for formant display. Uploaded data is assumed to be in Hz and will be automatically converted.

Filter visualizations to the five basic vowels (case-insensitive).

Supported Formats:

  • CSV/TXT/XLSX: Must include F1, F2, vowel columns (time, duration optional)
  • WAV + TextGrid: Audio file with label file - Extracts formants from labeled vowel segments
  • WAV only: Audio file only (automatic formant extraction)

šŸ’” Auto Column Detection: Automatically recognizes various column names (e.g., f1, F1, first_formant, formant_1, etc.)

šŸ“– TextGrid Formant Extraction Method

How formants are extracted from WAV + TextGrid files:

  1. Audio Processing: WAV files are automatically resampled to 16kHz for optimal formant analysis
  2. Vowel Detection: Only vowel segments (ARPABET: AH, IY, etc. / IPA: ÉŖ, ŹŠ, ə, etc.) are analyzed
  3. Sampling Region: Formants are measured from the middle 25% of each vowel's duration (37.5% ~ 62.5% timepoints) to avoid coarticulation effects
  4. Averaging: 5 measurement points within the middle region are sampled and averaged for stability
  5. Quality Control: Outliers beyond Ā±3 standard deviations (per vowel type) are automatically removed
  6. Praat Settings: Formant extraction uses Burg's method with 5 formants, 5500Hz ceiling, 25ms window, 10ms time step

šŸ“Š Result: Each vowel segment yields one F1/F2 measurement pair (averaged and rounded to 1 decimal place)