---
title: Voices & speaking models
description: Choose from roughly 50 high-quality voices, then pick a speaking model tuned for speed, stability, or expressiveness.
canonical: https://chatwisp.ai/knowledge-base/ai-interviewer/voices-and-models
updated: 2026-09-19
---

# Voices & speaking models

# Voices & speaking models

Your interviewer can speak in roughly **50 professional voices** — a range of genders, accents, and vibes from warm and soothing to energetic and authoritative.

## Pick a voice

Search and filter by name, gender, or personality, then **preview any voice** with one tap before you commit. If your workspace connects a custom voice provider, your own created voices appear here too.

The voice you choose is independent of which persona you pick, and it **sticks** — switching personas, adjusting dials, or tweaking other settings will not change it. Your voice only changes when you deliberately pick a new one.

## Choose your speaking model

Three speech engines are available, each tuned for a different priority:

- **Turbo (the default)** — the fastest, built for live, real-time conversation and instant dictation. The right choice for almost every form.
- **Multilingual (most stable)** — a polished, narration-grade engine. Best when the voice should sound more deliberate (long welcome messages, storytelling).
- **Expressive (most expressive)** — the most emotionally nuanced, with broad language support. Best for entertainment, quizzes, and brand experiences where character shines.

If a model is not available on your account, ChatWisp automatically falls back to a working one while keeping your chosen voice — your form never goes silent.

## How it works for respondents

The interviewer **speaks your questions aloud** in your chosen voice. Respondents **speak back**, and their words are transcribed live. Modern browsers support voice with no setup. Respondents can **barge in** — start talking mid-question to interrupt — and **live captions** show every word for clarity and accessibility.

## Common questions

### Will the voice I picked stay picked?
Yes. Once you choose a voice explicitly it becomes sticky, and only the voice picker changes it after that. Switching character archetypes will not silently swap it out from under you.

### Which voice should I use for a non-English form?
Pick one that speaks the language. The gallery filters by language and flags voices that are locked to one, so you can see at a glance whether a voice can carry the languages your form offers.

### Do premium voices cost more?
Enhanced voices consume more voice credits per response than the standard ones, and the gallery labels each with its multiplier before you choose. The default voice is the most economical option on every plan.

### What do respondents hear if a voice is unavailable?
The form falls back rather than going silent: it tries the next best model, then the next best voice, so the interview always has something to speak with.

## Related

[Personas & personality dials](/knowledge-base/ai-interviewer/personas) · [What your respondents experience](/knowledge-base/account-team/respondent-experience)