Text-to-speech services
Text-to-speech turns your bot's replies into the voice the caller hears. You select the service while creating a bot connection, and can change it later on the connection's Settings tab.
Set speech-to-text first. The language you select there filters the voices offered here, so the list holds only voices that speak the caller's language.
Bots on Google Dialogflow are the exception: Dialogflow brings its own speech services, so there is nothing to set.
The following table lists what each service asks you to set.
| Service | What you set |
|---|---|
| Microsoft | Voice |
| Model and voice | |
| Amazon | Engine and voice |
| Deepgram | Model, language, and voice, plus a data-sharing choice |
| Soniox | Model, language, and voice |
| ElevenLabs | Model and voice, by name or ID |
| Deepdub | Model and voice, by name or ID |
| Inworld | Model |
| Custom | Whatever your own provider needs |
The 'Text-to-speech provider' list offers the services AudioCodes has pre-deployed, plus any speech provider you have added yourself. Which region handles a pre-deployed service depends on your Live Hub region. See Global region support.