agentic-speech

route text to speech across elevenlabs, openai, and cartesia

the problem

each voice api speaks its own dialect. voice work locks to one vendor fast.

what i built

speech takes one text for all three. change voices without rewriting code.