What each measurement means:

• F1 (First Formant, Hz): Primarily indicates vowel height (vertical tongue position). High F1 = open vowels (अ "a", आ "ā"); Low F1 = closed vowels (इ "i", उ "u"). Mathematically, the lowest resonance frequency of the vocal tract.
• F2 (Second Formant, Hz): Primarily indicates vowel backness (front-back tongue position). High F2 = front vowels (इ "i", ई "ī"); Low F2 = back vowels (उ "u", ऊ "ū"). The second-lowest resonance frequency.
• F3 (Third Formant, Hz): Indicates finer details of articulation and consonant-vowel transitions. Higher values typically found in fricatives and consonants. The third-lowest resonance frequency of the vocal tract.
• Pitch (F0, Hz): Fundamental frequency—the rate of vocal cord vibration. Higher pitch = faster vibration (typically female speakers); Lower pitch = slower vibration (typically male speakers). Ranges ~80-250 Hz for adult speakers.
• Duration (milliseconds): Length of the phoneme. Vowels typically 100-300ms; consonants typically 50-200ms depending on manner of articulation.
• Spectral Centroid (Hz): Weighted mean frequency of the spectrum—indicates the "brightness" or "center of mass" of the sound. Higher centroid = brighter sound with more high-frequency energy (typical for fricatives like सु); Lower centroid = duller sound dominated by low frequencies (typical for vowels like ऊ). Ranges ~1000-4000 Hz for speech.
• Spectral Rolloff 95% (Hz): The frequency threshold containing 95% of the total sound energy. Marks where most of the acoustic energy is concentrated. Lower rolloff = energy concentrated in lower frequencies; Higher rolloff = energy spread across higher frequencies. Useful for distinguishing vowels from consonants.

Color Coding: Blue gradients show relative values within formant/spectral columns (lighter = lower, darker = higher); Red gradients show relative pitch values. This helps spot patterns across phonemes.

Quick visual comparison: Full-width cards showing Waveform and Pitch (left), Frequency Probability Distribution (right) for each varna. Uses the same filter as the Acoustic Features Summary table above.

F1 vs F2 formant space for each primary vowel. Each point represents one utterance. Ellipses show ±1 standard deviation clusters.

Pitch (F0) distribution by gender for each primary vowel. Box plots show median (black line), mean (yellow diamond), and quartiles.

Extended vowel inventory: ā, ī, ū, e, ai, o, au, ṛ. Same F1/F2 formant space visualization as primary vowels.

Pitch (F0) distribution by gender for each secondary vowel.

Articulatory variety: ka (velar), ca (palatal), ṭa (retroflex), ta (dental), pa (labial), ha (glottal), and ri, su (approximant/fricative). Formants reflect the inherent vowel component (a) or vowel context. These show consonantal acoustic properties and place-of-articulation distinctions.

Pitch (F0) distribution by gender for select vyanjanas.