Soniox is already supported as a transcriber provider. They've since launched

a Text-to-Speech product, and it would be useful to select it as a voice

provider too.

Why it's a good fit:

- 60+ languages with a single voice set, which is helpful for non-English and

multilingual agents. We run a Japanese-language agent and already use Soniox

STT, so having both sides from one provider would simplify things.

- Streaming synthesis over WebSocket, with audio starting before the full

sentence is available — suited to real-time voice.

- Competitively priced at $0.70/hour of generated speech.