Skip to content

use case

verified July 2026

TTS for phone surveys and research calls.

Survey research lives and dies on consistency — the same question, delivered the same way, ten thousand times. That is precisely what synthesis does that human panels cannot, and the field-window traffic shape is precisely what burst lines price sanely.

01

Why synthesis fits research

  • Zero interviewer effect: every respondent hears the identical stimulus, wave after wave.
  • A questionnaire re-versioned in 23 languages from the same voice keeps cross-language waves comparable.
  • Unmetered characters mean piloting five phrasings costs nothing but the calls themselves.

02

The field window, priced

A small pilot is cheaper on a meter — take the meter for pilots, honestly. The full field window on burst days is where the arithmetic flips, and the quarterly budget becomes a known number.

Table 1 — A quarterly wave, fielded in ten days

WaveMinutesOn a $30–$100 meterOn Gandr lines
Pilot, 1 line1,000$30 – $100$150
Field, 1 line + burst25,000$750 – $2,500$150 + burst days

03

Completion is a latency problem

Respondents abandon at silences. First audio inside the turn budget keeps completion rates from quietly becoming a synthesis-latency artifact — and the figure is published, so your methods section can cite it.

See also

Related sheets.

Thirty minutes, your production script, the live latency readout — measured in front of you.

Bring this use case to the call