Clarity
How much usable speech is present and whether the signal is easy to distinguish from silence or noise.
Method
What the voice report measures.
The report describes the delivery of one short recording. It is designed for speaking practice, not personality judgment, identity analysis, hiring, diagnosis, or any high-stakes decision.
From recording to report
- Local measurements. Your browser measures duration, loudness, silence, clipping, vocal activity, pitch movement, and related acoustic cues.
- Speech transcription. The recording is sent in one request to Cloudflare Workers AI so the app can estimate word count and speaking pace.
- Fixed scoring rules. A versioned scoring rubric combines those measurements into delivery and possible listener-impression scores.
- One next action. The report highlights a strength and suggests one change to try in another take.
The scores
Pace
Estimated words per minute and how close the take is to a comfortable conversational range.
Stability
Consistency of loudness and vocal delivery across the short sample.
Energy
The strength of the vocal signal, adjusted for recording quality and clipping.
Expressiveness
Variation in pitch and delivery rather than a judgment about emotion or personality.
Listener impression
Possible perceptions such as warmth or confidence, inferred from delivery cues and shown with a confidence level.
What can change the result
A different microphone, room, distance, background noise, sentence, accent, language, or speaking style can change the measurements. Scores from two recordings are most useful when the device and environment stay similar.
The report is intentionally limited. It cannot reveal character, truthfulness, attractiveness, health, gender identity, ethnicity, or professional ability.