Voice and TTS settings

Purpose

The Voice Config page controls how the assistant sounds on phone calls: the text-to-speech provider, the specific voice, and delivery tuning such as speed. A second tab holds an optional pre-recorded welcome audio clip.

Prerequisites

A tenant role with voice management permission; members without it do not see Voice Config in the sidebar.
Saving here changes the live phone voice for the tenant, so treat Save as a production change.

Steps

Step 1

voice-and-tts-settings, step-01

Action: Click Voice Config under Voice & Messaging in the sidebar.

What you should see: The Voice Configuration page opens on the Voice Settings tab. It shows the TTS Provider selector (ElevenLabs in the demo tenant), a Voice picker with a small preview play button beside it, and a Speed slider. The Save button stays disabled until a voice is chosen; when the tenant already has a voice set, it opens enabled. Platform admins see extra controls that regular members do not: a Model selector, Stability and Similarity Boost sliders (ElevenLabs only), and a Context Inbound (Asterisk) field above the provider. See Troubleshooting.

Step 2

voice-and-tts-settings, step-02

Action: Open the TTS Provider dropdown to see the available engines. Close it without changing the value.

What you should see: The dropdown lists the supported providers, ElevenLabs and Cartesia. Switching provider clears the current voice selection. For platform admins, the model selection also clears and the provider-specific sliders change (Stability and Similarity Boost apply to ElevenLabs only); regular members only ever see the Speed slider.

Step 3

voice-and-tts-settings, step-03

Action: Open the Voice picker.

What you should see: A searchable list of voices for the selected provider appears, each with a short character description such as “Bella - Professional, Bright, Warm”. The demo tenant’s ElevenLabs catalog offers 28 voices.

Step 4

voice-and-tts-settings, step-04

Action: Select a voice from the list. Do not click Save unless you intend to change the live assistant.

What you should see: The selected voice fills the picker with a clear (x) control. The preview play button next to it becomes active so you can hear a sample, and the Save button enables. Nothing is applied until you click Save. Leaving the page discards the selection.

Step 5

voice-and-tts-settings, step-05

Action: Switch to the Welcome Audio tab.

What you should see: The tab contains a single Upload Welcome Audio button that accepts an audio file. When a clip is present, it plays at the start of inbound calls in place of synthesized speech for the greeting.

Troubleshooting

Save stays disabled: it requires a voice to be selected. Pick a voice first, even if you only changed a slider.
The preview button is greyed out: it activates only after a voice is selected in the picker.
Your voice and model choices vanished after switching provider: this is intentional; each provider has its own catalog, so re-pick both.
You do not see the Model selector, the Stability slider, or the Asterisk context field: those controls are platform-admin only, and the two sliders also require the ElevenLabs provider.

Notes

Route: /settings/voice, sidebar item “Voice Config” under Voice & Messaging. Source: pages/settings/VoiceConfigPage.jsx, which renders the shared components/common/VoiceConfigPanel.jsx.
The sidebar item is gated by the voice.manage permission (platform admins always see it). This is a different gate from the rest of the settings pages, which mostly use tenants.settings.manage.
This page applies one voice tenant-wide for en-US; per-language voice overrides exist in the shared panel but are not enabled on this page.
Captures were taken in the Demo Auto Shop tenant as a regular Owner, which is why only the Speed slider and no Model or Stability controls appear. The voice selected in step 4 was not saved; the tenant’s stored settings were left untouched.