Frazo
Frazo turns live speech into translated audio for listeners. The product combines an operator console, a simple audience experience, and a relay service that keeps translation streams moving during a live session.
Separate listening from operating
The audience needs to follow the speaker while an operator manages the session and available languages. I separated those interfaces: listeners choose a translation and receive captions and audio in the browser, while session controls stay with the operator.
Keep audio delivery separate from the interface
Live translation needs continuous audio delivery as listeners join the session. I separated the audio relay from the viewer and operator tools, using WebSockets to distribute streams. This gives audio delivery its own service and leaves each interface responsible for its part of the session.