Spectral Analysis of Vowels

acoustics spectrum frequency analysis spectrograph
Last updated: 2025-01-19

Spectral Analysis of Vowels

A basic tool of analysis of vowels is the sound spectrograph. To understand this instrument, it is necessary to delve deeper into the concepts of spectral analysis. The spectrum of the glottal airflow waveform has already been described in Chapter 5, and spectra of resonance tubes (acoustic filters) have been introduced. This section expands these concepts to general time-frequency transformations of sound sources and filters.

The Need for Spectral Analysis

Vowel production involves complex interactions between:

  • The time-varying glottal source
  • The frequency-selective vocal tract filter
  • The resulting acoustic output

Spectral analysis allows us to:

  1. Separate source from filter: Understand contributions of larynx vs. vocal tract
  2. Identify formant frequencies: Measure F₁, F₂, F₃, etc.
  3. Quantify formant bandwidths: Assess energy loss and resonance quality
  4. Track formant movements: Follow vowel changes over time
  5. Compare vowels: Classify and categorize vowel productions

Time-Domain vs. Frequency-Domain

Time-domain representation:

  • Shows how acoustic pressure varies over time
  • Waveform display
  • Reveals periodicity and temporal structure
  • Difficult to see spectral content directly

Frequency-domain representation:

  • Shows how acoustic energy is distributed across frequencies
  • Spectrum display
  • Reveals formant structure and harmonic content
  • Loses temporal detail (which instant in time)

Both representations contain the same information, just organized differently. Fourier transformation is the mathematical tool that converts between them.

Historical Context

The mathematical foundations of spectral analysis were laid by Joseph Fourier (1768-1830), who showed that any periodic waveform can be represented as a sum of sinusoids. This fundamental insight underlies all modern speech and voice analysis.

The development of the sound spectrograph in the 1940s revolutionized speech research by providing visual representations of spectral content over time. This tool became essential for:

  • Phonetic research
  • Voice analysis
  • Speech therapy
  • Forensic voice identification
  • Singing pedagogy

Components of Spectral Analysis

1. Fourier Transformation

Converting time-domain signals to frequency-domain spectra through mathematical decomposition into sinusoidal components.

2. Filtering

Selective amplification or attenuation of frequency ranges, both as a physical process (vocal tract) and as an analysis tool.

3. Spectrographic Display

Visual representation combining time, frequency, and amplitude information in a single display.

Applications in Voice Science

Spectral analysis is used to:

In Research:

  • Study formant frequency patterns across vowels
  • Investigate source-filter interaction
  • Measure acoustic effects of articulatory changes
  • Analyze voice quality differences

In Clinical Practice:

  • Assess voice disorders through spectral characteristics
  • Document treatment outcomes
  • Provide visual feedback in therapy
  • Identify pathological acoustic signatures

In Pedagogy:

  • Teach vowel modification to singers
  • Provide real-time feedback on formant frequencies
  • Demonstrate effects of articulatory changes
  • Train speech and voice students

Overview of Topics

This section covers three main areas:

Fourier Transformations:

  • Principles of time-frequency transformation
  • Relationship between waveform duration and bandwidth
  • Different types of signals and their spectra
  • Practical implications for voice analysis

Filters and Filtering:

  • Types of filters (low-pass, high-pass, band-pass, band-reject)
  • Source-filter interaction in vowel production
  • Spectral multiplication and convolution
  • Filter bandwidth and selectivity

Sound Spectrograph:

  • Operating principles
  • Wideband vs. narrowband analysis
  • Reading and interpreting spectrograms
  • Applications in voice analysis

Spectral Analysis Workflow

A typical vowel analysis involves:

  1. Signal acquisition: Recording voice with appropriate quality
  2. Preprocessing: Removing noise, normalizing levels
  3. Transformation: Converting to frequency domain
  4. Feature extraction: Identifying formants, measuring bandwidths
  5. Interpretation: Relating acoustic measures to production and perception
  6. Documentation: Creating visual displays and reports

Summary

Spectral analysis provides essential tools for understanding vowel production and perception. By transforming time-domain signals into frequency-domain representations, we can separate the contributions of the glottal source and vocal tract filter, identify formant frequencies and bandwidths, and track changes over time. The sound spectrograph, based on principles of Fourier transformation and filtering, provides a powerful visual representation of vowel acoustics that has become indispensable in voice science research, clinical practice, and pedagogy.

Key Takeaways

  • ✅ Spectral analysis transforms time-domain signals to frequency-domain representations
  • ✅ Fourier transformation is the mathematical foundation for spectral analysis
  • ✅ Time and frequency representations contain the same information organized differently
  • ✅ The sound spectrograph combines time, frequency, and amplitude in visual displays
  • ✅ Spectral analysis enables separation of source and filter contributions
  • ✅ Applications span research, clinical practice, and vocal pedagogy

Further Reading

  1. Pickett, J. M. (1980). The sounds of speech communication. Baltimore: University Park Press.
  2. Borden, G. J., & Harris, K. S. (1980). Speech science primer: Physiology, acoustics and perception of speech. Baltimore: Williams & Wilkins.
  3. Baken, R. J., & Daniloff, R. G. (1991). Readings in clinical spectrography of speech. San Diego: Singular Publishing Group.
Back to Chapter 06