3. Acoustic Analysis

Enable and interpret acoustic overlays

Overview

Ozen-web can display multiple acoustic features overlaid on the spectrogram:

Pitch (F0) — Fundamental frequency (blue line and dots)
Formants (F1-F4) — Resonance frequencies (red dots)
Intensity — Loudness over time (green line)
HNR — Voice quality (harmonicity)
CoG — Spectral center of gravity
Spectral Tilt — High/low frequency balance
A1-P0 — Nasal formant measure

In this section, you’ll learn to enable these overlays and interpret what they show.

The Overlay Controls

Look for the overlay checkboxes in the interface (usually in a sidebar or below the spectrogram).

Enabling Pitch

Pitch (fundamental frequency) is the most commonly used overlay.

Check the “Pitch” checkbox
Wait 1-2 seconds for computation (first time only)
A blue line with dots appears on the spectrogram

Spectrogram with pitch overlay
Each dot represents one pitch measurement
Vertical position = frequency (Hz)

Interpreting the Pitch Overlay

Higher on screen = higher pitch
Lower on screen = lower pitch
Gaps in the line = unvoiced sounds ([s], [f], [t], [k], etc., or silence) or pitch detection failures
Smooth contours = steady pitch (often vowels)
Sharp movements = pitch changes (intonation / tone)

Try this: Record yourself saying “really?!” and watch the pitch track follow your intonation.

Note

Pitch is only detected in voiced segments. Consonants like [s], [f], [θ] won’t show pitch values.

Enabling Formants

Formants show the resonance frequencies that distinguish vowels.

Check the “Formants” checkbox
Red dots appear on the spectrogram

Spectrogram with formants overlay
Four vertical series of dots represent F1 (lowest), F2, F3, F4 (highest)

Interpreting the Formant Overlay

Formants reveal vowel identity:

Vowel	Appearance on Spectrogram
[i] peel	F1 low, F2 very high
[ʊ] pull	F1 low (but not as low), F2 low
[ɔ] Paul	F1 high, F2 low
[æ] pal	F1 high, F2 high
[ɜ˞] pearl	F1 mid, F2 high, F3 low

Try this: Find a vowel in your audio, place the cursor in the middle, and read F1 and F2 values in the values panel. Compare to the table above.

Tip

F1 roughly corresponds to tongue height (high = low F1, low = high F1). F2 roughly corresponds to tongue frontness (front = high F2, back = low F2).

Enabling Intensity

Intensity shows loudness over time.

Check the “Intensity” checkbox
A green line appears, usually in the upper part of the spectrogram

Spectrogram with intensity overlay
Height corresponds to decibels (dB)

Interpreting Intensity

Higher = louder segments (stressed syllables, vowels)
Lower = quieter segments (unstressed syllables, consonants)
Drops to bottom = silence

Try this: Say a sentence with strong stress on one word. The intensity overlay will peak on the stressed syllable.

Enabling Advanced Overlays

HNR (Harmonics-to-Noise Ratio)

Measures voice quality:

High HNR (>10 dB) = clear, modal voice
Low HNR (<5 dB) = breathy or creaky voice
Very low HNR = whisper or noise

Check the “HNR” checkbox
Orange/yellow line appears
Higher = more harmonic (less noisy)

CoG (Center of Gravity)

Measures spectral balance:

High CoG (>5000 Hz) = sibilants (s, sh), fricatives
Low CoG (<3000 Hz) = vowels, sonorants

Useful for distinguishing fricatives. To really see how fricatives are different, increase “Max Freq.” to 10,000, and record the word chefs.

Spectral Tilt

Measures high vs. low frequency emphasis:

Positive tilt = more energy in low frequencies (vowels)
Negative tilt = more energy in high frequencies (fricatives)

A1-P0

A measure related to nasality and open quotient. Useful for advanced research.

Note

Advanced overlays (HNR, CoG, Spectral Tilt, A1-P0) are primarily useful for specialized phonetic research. Most users only need Pitch, Formants, and Intensity.

Viewing All Overlays Together

You can enable multiple overlays simultaneously:

Warning

Enabling too many overlays at once can make the display cluttered. Start with Pitch + Formants for most tasks.

Reading Values at the Cursor

The values panel shows precise measurements at the cursor:

Enable the overlays you’re interested in (e.g., Pitch and Formants)
Place the cursor in a vowel
Read the values panel:
- Time: 0.452 s
- Freq: 1234 Hz (where you clicked)
- Pitch: 234 Hz
- Intensity: 68 dB
- F1: 523 Hz
- F2: 1987 Hz
- F3: 2743 Hz
- F4: 3543 Hz
Values panel with measurements

Tip

You can collect these measurements systematically using data points (covered in section 5).

Customizing Analysis Settings

You can adjust analysis parameters via the settings panel (if available) or config.yaml:

Pitch settings:

pitchFloor: Minimum pitch to detect (default: 75 Hz)
pitchCeiling: Maximum pitch to detect (default: 600 Hz)

Formant settings:

maxFormants: Number of formants to track (default: 5)
maxFormantFrequency: Analysis ceiling (default: 5500 Hz for female voices, 5000 Hz for males)

See Configuration Reference for all options.

Changing Spectrogram Max Frequency

You may want to adjust the vertical range of the spectrogram:

Find the Max Frequency dropdown (usually in toolbar)
Choose a ceiling:
- 5 kHz — Good for speech (default)
- 7.5 kHz — Includes higher formants
- 10 kHz — Full range, useful for fricatives
The spectrogram y-axis rescales

Tip

For most speech analysis, 5 kHz is sufficient. Use 7.5-10 kHz if studying sibilants or children’s voices.

Practice Exercises

Enable Pitch only
- Find a vowel
- Place cursor in the middle
- Note the pitch value
Enable Formants
- Identify F1 and F2 for the same vowel
- Try to guess the vowel based on formants (use the table above)
Enable Intensity
- Find the loudest part of your audio
- Compare intensity values across different sounds
Enable all overlays
- Observe how different overlays align
- Notice which features co-vary (e.g., intensity and voicing)

Troubleshooting

Pitch overlay doesn’t appear:

Ensure WASM backend is loaded (check backend selector)
Try enabling/disabling the checkbox
Check browser console (F12) for errors

Pitch values seem wrong:

Adjust pitchFloor and pitchCeiling for your speaker
Male voices: Try 50-300 Hz range
Female voices: Try 100-500 Hz range
Children: Try 150-600 Hz range

Formants are missing or erratic:

Ensure you’re looking at a vowel (not a consonant)
Try adjusting maxFormantFrequency (5500 for females, 5000 for males)
Zoom in to see individual formant dots more clearly

Too many overlays, display is cluttered:

Disable overlays you’re not actively using
Use the values panel to read measurements instead of viewing all overlays

Overlays don’t update when zooming:

This is expected for files >60s — analysis recomputes when zoomed
Wait 1-2 seconds after zooming for the update

What’s Next?

Now that you can visualize acoustic features, let’s learn how to create annotations to mark boundaries and add labels.

Next: 4. Annotations →

Navigation: ← Previous: Exploring Audio | Tutorial Overview | Next: Annotations →

--- title: "3. Acoustic Analysis" subtitle: "Enable and interpret acoustic overlays" --- ## Overview Ozen-web can display multiple acoustic features overlaid on the spectrogram: - **Pitch (F0)** — Fundamental frequency (blue line and dots) - **Formants (F1-F4)** — Resonance frequencies (red dots) - **Intensity** — Loudness over time (green line) - **HNR** — Voice quality (harmonicity) - **CoG** — Spectral center of gravity - **Spectral Tilt** — High/low frequency balance - **A1-P0** — Nasal formant measure In this section, you'll learn to enable these overlays and interpret what they show. ## The Overlay Controls Look for the **overlay checkboxes** in the interface (usually in a sidebar or below the spectrogram). ![Overlay toggle controls](../screenshots/tutorial-03-overlay-checkboxes.png){.screenshot} ## Enabling Pitch Pitch (fundamental frequency) is the most commonly used overlay. ::: {.tutorial-steps} 1. **Check the "Pitch" checkbox** 2. Wait 1-2 seconds for computation (first time only) 3. A **blue line with dots** appears on the spectrogram ![Spectrogram with pitch overlay](../screenshots/tutorial-03-pitch-overlay.png){.screenshot} 4. Each dot represents one pitch measurement 5. Vertical position = frequency (Hz) ::: ### Interpreting the Pitch Overlay - **Higher on screen** = higher pitch - **Lower on screen** = lower pitch - **Gaps in the line** = unvoiced sounds ([s], [f], [t], [k], etc., or silence) or pitch detection failures - **Smooth contours** = steady pitch (often vowels) - **Sharp movements** = pitch changes (intonation / tone) ::: {.callout-try} **Try this:** Record yourself saying "really?!" and watch the pitch track follow your intonation. ::: ::: {.callout-note} Pitch is only detected in **voiced** segments. Consonants like [s], [f], [θ] won't show pitch values. ::: ## Enabling Formants Formants show the resonance frequencies that distinguish vowels. ::: {.tutorial-steps} 1. **Check the "Formants" checkbox** 2. **Red dots** appear on the spectrogram ![Spectrogram with formants overlay](../screenshots/tutorial-03-formant-overlay.png){.screenshot} 3. Four vertical series of dots represent F1 (lowest), F2, F3, F4 (highest) ::: ### Interpreting the Formant Overlay Formants reveal vowel identity: | Vowel | Appearance on Spectrogram | |-------|---------|---------|---------------------------| | [i] _peel_ | F1 low, F2 very high | | [ʊ] _pull_ | F1 low (but not as low), F2 low | | [ɔ] _Paul_ | F1 high, F2 low | | [æ] _pal_ | F1 high, F2 high | | [ɜ˞] _pearl_ | F1 mid, F2 high, F3 low | ::: {.callout-try} **Try this:** Find a vowel in your audio, place the cursor in the middle, and read F1 and F2 values in the values panel. Compare to the table above. ::: ::: {.callout-tip} F1 roughly corresponds to **tongue height** (high = low F1, low = high F1). F2 roughly corresponds to **tongue frontness** (front = high F2, back = low F2). ::: ## Enabling Intensity Intensity shows loudness over time. ::: {.tutorial-steps} 1. **Check the "Intensity" checkbox** 2. A **green line** appears, usually in the upper part of the spectrogram ![Spectrogram with intensity overlay](../screenshots/intensity-overlay.png){.screenshot} 3. Height corresponds to decibels (dB) ::: ### Interpreting Intensity - **Higher** = louder segments (stressed syllables, vowels) - **Lower** = quieter segments (unstressed syllables, consonants) - **Drops to bottom** = silence ::: {.callout-try} **Try this:** Say a sentence with strong stress on one word. The intensity overlay will peak on the stressed syllable. ::: ## Enabling Advanced Overlays ### HNR (Harmonics-to-Noise Ratio) Measures voice quality: - **High HNR** (>10 dB) = clear, modal voice - **Low HNR** (<5 dB) = breathy or creaky voice - **Very low HNR** = whisper or noise ::: {.tutorial-steps} 1. Check the "HNR" checkbox 2. Orange/yellow line appears 3. Higher = more harmonic (less noisy) ::: ### CoG (Center of Gravity) Measures spectral balance: - **High CoG** (>5000 Hz) = sibilants (s, sh), fricatives - **Low CoG** (<3000 Hz) = vowels, sonorants Useful for distinguishing fricatives. To really see how fricatives are different, increase "Max Freq." to 10,000, and record the word _chefs_. ### Spectral Tilt Measures high vs. low frequency emphasis: - **Positive tilt** = more energy in low frequencies (vowels) - **Negative tilt** = more energy in high frequencies (fricatives) ### A1-P0 A measure related to nasality and open quotient. Useful for advanced research. ::: {.callout-note} Advanced overlays (HNR, CoG, Spectral Tilt, A1-P0) are primarily useful for specialized phonetic research. Most users only need Pitch, Formants, and Intensity. ::: ## Viewing All Overlays Together You can enable multiple overlays simultaneously: ![All overlays enabled](../screenshots/tutorial-03-all-overlays.png){.screenshot} ::: {.callout-warning} Enabling too many overlays at once can make the display cluttered. Start with Pitch + Formants for most tasks. ::: ## Reading Values at the Cursor The **values panel** shows precise measurements at the cursor: ::: {.tutorial-steps} 1. Enable the overlays you're interested in (e.g., Pitch and Formants) 2. Place the cursor in a vowel 3. Read the values panel: - **Time:** 0.452 s - **Freq:** 1234 Hz (where you clicked) - **Pitch:** 234 Hz - **Intensity:** 68 dB - **F1:** 523 Hz - **F2:** 1987 Hz - **F3:** 2743 Hz - **F4:** 3543 Hz ![Values panel with measurements](../screenshots/tutorial-05-datapoint-values.png){.screenshot} ::: ::: {.callout-tip} You can collect these measurements systematically using **data points** (covered in section 5). ::: ## Customizing Analysis Settings You can adjust analysis parameters via the settings panel (if available) or `config.yaml`: **Pitch settings:** - `pitchFloor`: Minimum pitch to detect (default: 75 Hz) - `pitchCeiling`: Maximum pitch to detect (default: 600 Hz) **Formant settings:** - `maxFormants`: Number of formants to track (default: 5) - `maxFormantFrequency`: Analysis ceiling (default: 5500 Hz for female voices, 5000 Hz for males) See [Configuration Reference](../reference/configuration.html) for all options. ## Changing Spectrogram Max Frequency You may want to adjust the vertical range of the spectrogram: ::: {.tutorial-steps} 1. Find the **Max Frequency** dropdown (usually in toolbar) 2. Choose a ceiling: - **5 kHz** — Good for speech (default) - **7.5 kHz** — Includes higher formants - **10 kHz** — Full range, useful for fricatives 3. The spectrogram y-axis rescales ::: ::: {.callout-tip} For most speech analysis, **5 kHz** is sufficient. Use 7.5-10 kHz if studying sibilants or children's voices. ::: ## Practice Exercises 1. **Enable Pitch only** - Find a vowel - Place cursor in the middle - Note the pitch value 2. **Enable Formants** - Identify F1 and F2 for the same vowel - Try to guess the vowel based on formants (use the table above) 3. **Enable Intensity** - Find the loudest part of your audio - Compare intensity values across different sounds 4. **Enable all overlays** - Observe how different overlays align - Notice which features co-vary (e.g., intensity and voicing) ## Troubleshooting **Pitch overlay doesn't appear:** - Ensure WASM backend is loaded (check backend selector) - Try enabling/disabling the checkbox - Check browser console (F12) for errors **Pitch values seem wrong:** - Adjust `pitchFloor` and `pitchCeiling` for your speaker - Male voices: Try 50-300 Hz range - Female voices: Try 100-500 Hz range - Children: Try 150-600 Hz range **Formants are missing or erratic:** - Ensure you're looking at a vowel (not a consonant) - Try adjusting `maxFormantFrequency` (5500 for females, 5000 for males) - Zoom in to see individual formant dots more clearly **Too many overlays, display is cluttered:** - Disable overlays you're not actively using - Use the values panel to read measurements instead of viewing all overlays **Overlays don't update when zooming:** - This is expected for files >60s — analysis recomputes when zoomed - Wait 1-2 seconds after zooming for the update ## What's Next? Now that you can visualize acoustic features, let's learn how to create annotations to mark boundaries and add labels. **Next:** [4. Annotations](04-annotations.html) → --- **Navigation:** [← Previous: Exploring Audio](02-exploring-audio.html) | [Tutorial Overview](index.html) | [Next: Annotations →](04-annotations.html)