Choosing a text-to-speech provider

The synthesizer turns the agent’s reply into audio. Latency and language are the primary selection factors. Always enable stream: true.

Provider	Best for	Notes
ElevenLabs	Natural English voices	Bolna default; Turbo models are fastest
Cartesia	Lowest latency English	Very fast time-to-first-audio
Azure TTS	Multilingual, enterprise	Strong across many languages
AWS Polly	Cost-sensitive workloads	Lower cost; neural voices available
Deepgram Aura	English	Fast and accurate
Rime	Natural conversational English	Good prosody for dialogue
Sarvam	Indian languages	Native voices for Hindi, Tamil, Telugu, and more
Smallest AI	Ultra-low latency	Optimized for real-time
Maya	Indian languages	11 languages, 2 voices, mid-call language switching without reconnecting
Kalpa Labs	English and Hinglish	Code-switched Hindi and English from one model, no language parameter
Soniox	One provider for STT and TTS	Real-time `tts-rt-v2`; voice cloning, speed and silence controls; same `SONIOX` key as the transcriber
OpenAI	Simple English voices	`tts-1` with OpenAI’s own voice names; not browsable via the voice API

Key settings: stream: true (always enable), buffer_size (100–250 chars typical), audio_format (must match your telephony provider). For the exact provider and model values each takes, see Create Agent.

gemini appears as a synthesizer provider but cannot currently be used — no Gemini TTS model resolves.

Next steps

Text-to-speech providers

Setup guides for every voice provider

Clone a voice

Create a custom voice for your agent

Choose telephony

Pick how calls reach your agent

Back to the decision guide

Compare all provider categories side by side

Telephony

AI models

No-code automation

Choose providers

Choosing a text-to-speech provider

Next steps

Text-to-speech providers

Clone a voice

Choose telephony

Back to the decision guide

​Next steps

Text-to-speech providers

Clone a voice

Choose telephony

Back to the decision guide

Next steps