Twilio Changelog | Nov. 18, 2026
Conversation Relay now supports speech-to-text and text-to-speech provider failover
You can now add backup speech-to-text (STT) and text-to-speech (TTS) providers to Conversation Relay. If a provider fails, the call switches to its backup and stays connected.
When the STT or TTS provider that a Conversation Relay call uses fails, Conversation Relay switches to a backup provider. The caller stays on the line, and your application keeps the same WebSocket connection. Failover is automatic, so your application doesn’t need to detect or react to the failure.
STT and TTS fail over independently, at call setup, or mid-call. The backup voice speaks any text that hasn’t been spoken yet, in order. Speech recognition restarts on the backup provider, and speech during the switch might be missed. To see the switch, subscribe to the config-updated event. Your application then receives a configUpdated info message that identifies the new provider.
To use failover, add a <Failover> element to your <ConversationRelay> TwiML. Specify the backup provider, voice, and model for each language, or use Twilio’s default backups.
Keep these limits in mind:
Failover happens once for TTS and once for STT per call. If the backup also fails, the call ends.
You set failover in TwiML. Configuration updates during a call can’t change it.
Some languages have no default backup, so you need to specify one.
For detailed setup instructions, examples, and default backups, see Provider failover for Conversation Relay.