Visualize Phonemes through Waveform and Spectrogram

Vowels /i ɛ a ʌ u/ etc.

  • Vowels are periodic sounds, i.e., you can find the same pattern repeating in the waveform.

  • They have clear formants -- which are dark bands in the spectrogram.

  • They are voiced, and you can always see pitch associated with vowels.

The waveform of the vowel /ɛ/

The spectrogram of the vowel /ɛ/

Nasals /n m ŋ/

  • Nasals are also voiced (with pitch)

  • They have lighter spectrum energy than vowels, sometimes with the high-frequency region missing.

  • The waveforms are also periodic, but the shapes of their cycles are smoother and have fewer fluctuations.

  • Their magnitude is much smaller than vowels.

Comparison between a vowel (left) and a nasal (right) (a)

Comparison between a nasal (left) and a vowel (right) (b)

Plosives /b p t d k g/

Voiceless plosives: /p t k/

  • silence + a burst + aspiration in waveform.

  • a vertical energy bar in the spectrogram.

  • no pitch.

The waveform and spectrogram of a word-initial /p/ sound

Voiced plosives /b d g/ are more subtle, sometimes all that you get in visualization is a burst.

/b/

Fricatives /h f v s z ʃ ʒ θ ð/

Voiceless fricatives: /h f s ʃ θ/

  1. Fricatives have aperiodic waveforms.

  2. The magnitude of the waveform builds up gradually (as opposed to a sudden burst).

  3. The spectral energy is concentrated on the high-frequency region.

  4. No pitch.

The waveform and spectrogram of /s/ (coming before a vowel)

Zooming in: the aperiodic waveforms of /s/

Resources on Youtube: