- Deepgram
- Cartesia
- Fish Audio
- Inworld
- Sarvam AI
- KugelAudio
- Murf
- Soniox
- Gradium
Models
Text to speech
Text-to-speech (TTS) models take written text and turn it into natural-sounding spoken audio. In SLNG voice agents this is the “Speak” step: the LLM sends back a response and the TTS model gives it a voice the caller hears. The voice’s quality and speed shape how the whole agent feels. A natural voice builds trust, and a fast one keeps the conversation flowing without awkward pauses.
TTS providers we support: