Kotoba Speech APIs Land on fal

Kotoba’s fal launch makes its low-latency voice stack much easier to test and integrate, especially for multilingual agents and realtime interfaces.

• –Speech-to-speech translation returns text and audio while conversations are still happening
• –Streaming STT exposes partial transcript updates for microphones, call systems, and voice agents
• –TTS supports streamed audio output and preset voices for responsive applications
• –File transcription adds optional word timestamps for recordings and offline workflows
• –fal’s usage-based pricing lowers experimentation friction, though cold-start latency and private-alpha access may limit production readiness