Vocal Analyzer Docs

Vocal Analyzer is an open-source tool for exploring the acoustic side of voice feminization, masculinization, or androgynization training. It turns a short recording into a set of research-grounded measurements — pitch, formants, voice quality, sibilants, and more — plus reference bands that show where the recording sits relative to published central tendencies for perceived-masculine and perceived-feminine voices.

These docs are for users of the tool: how to record well, what each number on the dashboard means, and where the limits of the analysis lie. If you are here to understand the internals, the developer docs under docs/ARCHITECTURE.md in the repo are a better entry point.

Where to start

• New to the tool? Work through Getting started, Recording tips, and Your first analysis.
• Just ran an analysis and want to know what the numbers mean? Read Interpreting your first result end to end — it walks a sample dashboard tile by tile.
• Curious about the science? Start with Voice and gender perception, then branch into the per-metric reference pages.
• Looking up one specific metric? Jump straight to its page in the sidebar under Metric reference.

What this site covers

• Tutorial: how to get a clean recording and read your first result.
• Explanation: why voice sounds "gendered" to listeners and how that maps to the measurements the tool computes.
• Reference: one page per metric. Each one states what the measure is, which published bands are used as overlays, and which recording conditions degrade it.
• How-to: a dashboard walkthrough, goal profiles, and the settings you can tweak from the sidebar.
• Methodology and caveats: confidence levels, measurement warnings, and the philosophy behind descriptive-vs-prescriptive reporting.
• Reference helpers: glossary, FAQ, troubleshooting.

A note on framing

Vocal Analyzer is a descriptive tool. It shows you where your recording sits against published central tendencies for perceived-masculine and perceived-feminine voices — it does not decide whether a voice sounds any particular way to any particular listener. Perception is a separate dimension (Dacakis et al. 2017). Treat the bands and numbers as information, not as targets or verdicts.

The tool is also not a voice teacher. If you are doing active voice training, a teacher can hear things that no acoustic measurement will catch. See Beyond the tool for pointers.