Soniox text-to-speech for Bolna AI voice agents

What is Soniox TTS used for in a Bolna voice agent?

Soniox text-to-speech turns your Bolna agent's replies into speech with its real-time tts-rt-v2 model, which covers 63 languages and 200 named voices from one model. Audio starts playing before the full sentence is ready, because Soniox streams speech as the reply is generated.

Pick it when you want one synthesizer across many languages, or one provider for both transcription and voice.

How do I set up Soniox TTS with Bolna?

1. Optional: add your SONIOX key under Settings → Integrations at platform.bolna.ai/providers. The same key serves the Soniox transcriber.
2. Open your agent in Agent Studio and go to the Languages tab to browse and preview Soniox voices, or build the agent through the Create Agent API.
3. Set the synthesizer provider to soniox and the model to tts-rt-v2.
4. Set both voice and voice_id to the voice name, for example Adrian. Both are required. For a cloned voice, put its id in voice_id.
5. Set language inside provider_config as a plain ISO code such as hi, not hi-IN. It is required.
6. Optional: set speed between 0.7 and 1.3, and reduce_silence to shorten pauses between sentences.

For every setting, see the Soniox TTS setup guide in the Bolna docs.

Which Soniox TTS models and voices does Bolna support?

Model	Status	Languages and voices
tts-rt-v2	Current real-time model, default on Bolna	63 languages, 200 voices
tts-rt-v1	Deprecated by Soniox	Use tts-rt-v2 on new agents

A few voices from the catalog: Adrian (male, deep and composed), Emma (female, British), Arjun (male, Indian English), Priya (female, Indian accent) and Divya (female, Tamil). Indian languages include Hindi, Bengali, Tamil, Telugu, Gujarati, Kannada, Malayalam, Marathi, Punjabi and Urdu.

What can you build with Soniox TTS and Bolna?

One provider for both sides of the call. Pair Soniox TTS with the Soniox transcriber so one SONIOX key covers speech-to-text and text-to-speech.

Multilingual agents with one synthesizer. Because one model covers all 63 languages, a multilingual agent keeps the same synthesizer on every language tab and only changes the voice.

A distinct brand voice. Clone a voice in your Soniox account and use it on your agents by passing its id.

Natural turn-taking. When a caller interrupts, the agent stops mid-sentence and drops the rest of the utterance, so it listens instead of talking over them.