Acoustic Measures of Voice Quality
A Living Reference
Jorge C. Lucero 
Welcome
A living reference. It grows chapter by chapter, and each release is deposited on Zenodo with its own versioned DOI. Pages marked Living draft are complete and sourced, and are revised as new sources are added. What changed in each release — including corrections to values published earlier — is recorded on the Release notes and errata page.
Acoustic Measures of Voice Quality is an open reference on the acoustic measures used to assess the human voice. It is written for speech-language pathologists and voice scientists. For each measure, it covers what the measure means, how it is computed, which reference values exist, and where the measure fails.
What makes it different
- Every reference value comes with its pipeline. A CPPS cutoff from one program does not apply to a value from another program, and an AVQI threshold validated in one language does not apply to another. Every normative table states the software, version, settings, speech task and language behind each value, and its compatibility class.
- Absence is declared, not filled. Where no published reference matches a pipeline, task or language, the table says so instead of borrowing a number.
- Disagreements are shown, not averaged. When sources conflict, including misprinted equations or inconsistent summary tables, the conflict is set out with both sources.
- Primary sources. Claims are traced to the original studies. Values are never taken from summary tables without saying so.
In this release
Foundations
- Signal Fundamentals: sampling, quantization, framing and windows, and the time, frequency and cepstral domains, with computed figures.
- Signal Typing: the Type 1–4 classification, and which measures are valid for which signals.
- Pipeline Dependence and Reference Values: why a value belongs to the software, settings, task and language that produced it, and how this reference reports reference values.
Periodicity and stability
- Jitter: does each cycle last as long as the one before it?
- Shimmer: is each cycle as large as the one before it?
Noise and harmonic structure
- Harmonic-to-Noise Ratio (HNR): how much of the sound is periodic at all?
- Cepstral Peak Prominence (CPP and CPPS): how strongly does a harmonic pattern stand out from the background?
Composite indices
- Acoustic Voice Quality Index (AVQI): how severe does the voice sound overall?
- Acoustic Breathiness Index (ABI): how breathy does it sound?
The jitter, shimmer and HNR chapters publish no cutoff: no reference value in the sources reviewed is compatible with a Praat-based pipeline, so the absence is declared. Each explains why, and what the published values from other systems can and cannot be used for.
Coming next
The complete reference is planned in five parts. The chapters below are not yet written.
- Foundations: recording protocol; F0 detection; voice tasks; perceptual anchors (GRBAS, CAPE-V).
- Periodicity and stability: F0 measures; voice breaks; perturbation caveats.
- Noise and harmonic structure: NHR; GNE; spectral tilt.
- Composite indices: the Dysphonia Severity Index (DSI); caveats shared by composite indices.
- Machine-learning-derived measures: a landscape chapter and two worked examples.
- Appendices: adjunct methods, reproducibility notes and a glossary.
Citing this reference
Cite the concept DOI, which always resolves to the latest release:
Lucero, J. C. (2026). Acoustic Measures of Voice Quality: A Living Reference. Zenodo. https://doi.org/10.5281/zenodo.22739049
To cite a specific release, use the versioned DOI listed on that release’s Zenodo record.