WebSpeech Live Voiceprint Formant & MFCC Spectrogram Analyzer (2026)

Fundamental Frequency (F0) is the rate at which your vocal folds vibrate (typically 85–180 Hz for adult males and 165–255 Hz for adult females), perceived as musical pitch. Formants (F1, F2, F3) are acoustic resonances created by the shape and length of your pharyngeal, oral, and nasal cavities—they determine which vowel sound is heard and uniquely characterize a speaker's physical vocal tract.

Human hearing resolves frequencies logarithmically rather than linearly—we distinguish 100 Hz from 200 Hz easily, but not 8,100 Hz from 8,200 Hz. MFCCs warp the FFT power spectrum onto the perceptual Mel scale, take the logarithm of filterbank energies, and apply a Discrete Cosine Transform (DCT) to decorrelate vocal tract shape (smooth spectral envelope) from glottal pitch.

Countermeasure models (such as ASVspoof RawNet2 and AASIST) inspect sub-band phase discontinuities, absence of natural glottal micro-jitter (0.2%–1.0%), uniform breath pauses, high-frequency neural vocoder checkerboard artifacts above 8 kHz, and secondary loudspeaker impulse convolutions caused by physical replay devices.

Laryngitis or nasal congestion swells the vocal folds (lowering F0 and increasing jitter/shimmer) and dampens nasal/pharyngeal formants. Modern x-vector and ECAPA-TDNN speaker embeddings focus on invariant craniofacial ratios (F3/F4 spacing and cepstral deltas) to minimize False Rejections during mild illness.

Never. All audio sampling, Hamming windowing, FFT spectral computation, and MFCC extraction run inside your browser's local WebAudio graph. Zero audio frames leave your device.