How to add a Text-to-Speech entry in VICIdial

Adding a TTS (text to speech) entry in VICIdial is quick once Cepstral is installed and the TTS system setting is enabled. The add form is deliberately short and asks only for what a valid record needs; the rest of the options appear after you save. This post walks through creating an entry and getting it to speak.

Confirm the prerequisites

Before you add anything, make sure Cepstral is installed and configured on the server and that the System Settings option for TTS is enabled. If either is missing the entry saves but never generates audio. Each server that runs TTS needs its own Cepstral channel license, voice, and save-to-file license.

Create the entry

Click ADD A NEW TTS ENTRY in the left menu of the admin. You must set the TTS ID and name or the submission is rejected.

1. TTS ID: the short unique identifier, two to fifty characters. Use only numbers, letters, and underscores, with no spaces or punctuation.
2. TTS Name: the descriptive label summarising the entry, two to one hundred characters.
3. Submit to save. The full set of options then appears on the modify screen.

Finish on the modify screen

After saving, modify the entry to set the rest:

• Active: turn the entry on or off.
• TTS Voice: the voice used for generation. The default is Allison-8kHz.
• TTS Text: the words sent to Cepstral. This is where you type what the caller hears.

Two more fields share the modify screen. The Admin User Group restricts which admin user groups can view the record; the default of --ALL-- lets any admin see it. TTS Voice is also here, and unless you change it the entry uses the default Allison-8kHz voice. None of these block the record from saving, so you can leave them at their defaults and refine later once you hear how the entry sounds.

How the entry turns into audio

When a call flow reaches the entry, its TTS Text is sent to Cepstral, an audio file is created, and that file plays to the caller. Reference the entry from an IVR (interactive voice response) or Call menu step so it fires at the right moment in the call. Because the audio is generated on demand, you can change the text and hear the new wording without re-recording anything; the next render picks up the edit.

If nothing plays, walk back through the prerequisites before touching the entry itself. The most common cause is Cepstral not being installed, or the System Settings TTS option being off, in which case the entry saves fine but never produces sound. After that, confirm the entry is set to Active and that the call flow actually points at the right TTS ID.

flowchart TD
  A["ADD A NEW TTS ENTRY"] --> B["Set TTS ID and Name"]
  B --> C["Submit to save"]
  C --> D["Modify entry"]
  D --> E["Set Active, Voice, Text"]
  E --> F{"Cepstral + setting on?"}
  F -->|"Yes"| G["Entry speaks on call"]
  F -->|"No"| H["No audio generated"]

If you are not sure what the ID and name fields control, the TTS ID and name breakdown goes deeper, and the audio and TTS guide frames where TTS fits among prompts and Recording format (WAV/MP3) choices. To skip the Cepstral setup entirely, VICIfast provisions a wired dialer in under 40 seconds, so see the plans.