How to Read a Spectrogram in Phonetics for Speech Analysis
Interpreting these visual representations of sound is a fundamental skill in acoustic phonetics. While software settings like window size and filter bandwidth drastically alter how a signal appears, the underlying physical properties of the vocal tract remain constant. Beginners often struggle because the same phoneme can look different depending on the speaker's age, gender, or the surrounding phonetic environment, known as coarticulation.
What are the primary visual components of a spectrogram, and how do they correspond to acoustic properties?
Spectrograms map time horizontally and frequency vertically, with energy intensity represented by grayscale shading. These visual patterns reflect acoustic properties, such as the 0-8000 Hz range typically used to capture human speech. Darker bands indicate high-energy resonance, while lighter areas represent lower intensity, allowing you to visualize complex sound waves.
Quick facts
| Feature | Visual Indicator | Typical Range |
|---|---|---|
| Time | Horizontal axis | 0–5 seconds |
| Frequency | Vertical axis | 0–8000 Hz |
| Intensity | Darkness/Color | 0–60 dB |
How can the fundamental frequency (F0) be identified on a wideband versus narrowband spectrogram?
You identify the fundamental frequency by observing the vertical striations representing glottal pulses in wideband displays, whereas narrowband spectrograms display individual horizontal harmonics.
| Method | Visual Feature | Accuracy |
|---|---|---|
| Wideband | Vertical glottal pulses | High temporal |
| Narrowband | Horizontal harmonics | High spectral |
| Wideband | 300 Hz filter | Broad resolution |
| Narrowband | 45 Hz filter | Fine resolution |
Measuring the vertical spacing between the base harmonic and the first overtone in a narrowband display provides the most precise calculation of the speaker's pitch.
What visual cues distinguish voiced sounds from voiceless sounds in a spectrogram?
Voiced sounds are characterized by a distinct "voice bar," a low-frequency band of energy typically located below 200 Hz, which signifies vocal fold vibration. In contrast, voiceless sounds lack this low-frequency energy and often show a sudden, diffuse pattern of high-frequency noise. For example, a voiced /z/ will show a clear voice bar throughout its duration, while a voiceless /s/ remains entirely silent in the low-frequency range.
An important nuance is that the voice bar may be faint or absent in recordings with significant high-pass filtering, making it difficult to detect even when vocal fold vibration is present. During the transition into a voiceless stop, the vocal folds may stop vibrating slightly before the closure is complete, which can lead to a deceptive visual signal of voicelessness.
How are the first three formants (F1, F2, and F3) located, and what do they indicate about vowel articulation?
The first three formants appear as dark, horizontal bands of energy. F1 is inversely related to tongue height, while F2 correlates with the front-back position of the tongue. F3 is sensitive to lip rounding and the position of the tongue blade. Many students assume that formants are single lines, but they are actually broad frequency regions representing the resonant cavities of the vocal tract. You must look for the center of these bands to track shifts accurately during vowel transitions.
What specific acoustic features characterize the production of stop consonants, such as voice onset time and burst release?
Stop consonants are identified by a silent gap followed by a sharp, vertical burst release. Voice onset time, the duration between the burst and the start of voicing, typically ranges from 20 to 100 milliseconds for voiceless aspirated stops in English, providing a clear marker for phonemic categorization.
How do fricatives appear on a spectrogram compared to nasal consonants?
Fricatives appear as broad, high-frequency bands of turbulent noise, often spanning several thousand hertz. Nasal consonants, by contrast, are defined by low-frequency resonances called nasal formants, usually below 1000 Hz, and visible anti-resonances that weaken the energy of higher formants. For instance, an /s/ shows intense energy above 4000 Hz, while an /m/ displays a strong, low-frequency pole near 250 Hz.
A common nuance is that the spectral shape of fricatives changes significantly based on the place of articulation; alveolar fricatives like /s/ show higher frequency energy than postalveolar ones like /ʃ/. Nasal consonants can sometimes be mistaken for vowels if the nasal murmur is short or the recording quality is poor, as both share a similar harmonic structure in the lower registers.
What is the difference between temporal resolution and frequency resolution in spectrogram settings, and why does it matter for phonetic analysis?
Temporal resolution focuses on precise timing of events, while frequency resolution prioritizes the separation of individual harmonic components. You must balance these settings because increasing one inherently decreases the other.
| Setting | Resolution | Trade-off |
|---|---|---|
| Wideband | High temporal | Low frequency |
| Narrowband | High frequency | Low temporal |
| 5 ms window | Fast tracking | Blurred harmonics |
| 40 ms window | Clear harmonics | Smearing time |
Select a wideband setting to analyze rapid transitions in consonants and a narrowband setting to measure stable pitch or harmonic spacing in vowels.
How can a researcher distinguish between liquid consonants and glides based on their spectral patterns?
Liquids and glides are distinguished by their formant transitions and the presence of steady-state energy, with liquids often showing more stable, lower-frequency F3 patterns.
- 1st Formant: Liquids like /l/ often exhibit a lower F1 compared to glides, reflecting the specific constriction of the tongue against the alveolar ridge.
- 3rd Formant: Glides, particularly /j/, show a dramatic upward movement of F3, whereas /l/ maintains a relatively consistent F3 position throughout the production.
- Spectral Energy: Liquids typically display a clearer, more defined F2 and F3 structure, while glides often appear more blurred due to their rapid, dynamic nature.
Common mistakes
- Ignoring the window size. Using a wideband setting for pitch analysis often obscures harmonics, so switch to narrowband when you need to see individual glottal pulses.
- Confusing noise with speech. High-frequency energy isn't always a fricative, as background white noise can mimic these patterns during recording.
- Overlooking coarticulation. Expecting a phoneme to look identical in every context leads to errors, as the influence of neighboring sounds constantly shifts formant transitions.
Refining Your Spectrogram Analysis Skills
Practice by recording your own voice and comparing the visual output to known phonetic charts. Consistent exposure to varied speech samples will help you recognize the subtle patterns that define human language.