Measure voice traits in generated phone audio, not the reference alone

A source recording's accent label, pitch, and conversational style do not guarantee those traits in reference-conditioned synthesis. Screen generated phone-band clips on matched phrases for pronunciation, pitch distribution, low-register outliers, and contour variation before selecting a voice. A spontaneous source may still yield flatter, lower speech than a read-speech source; the reverse can also occur. These measurements reject specific candidates but do not establish perceived naturalness or accent without calibrated listening or phonetic controls.

Applies when: choosing a voice reference for a narrowband conversational speech service. Keywords: TTS voice reference accent prosody conversational pitch phone audio validation