“I built live translated subtitles for video calls alone. I can talk about the cost, the limits, and calls you only half understand”
I am Sergei Skrylkov. I built Trippi alone: a Chrome and Edge extension that shows live captions of a Zoom, Meet or Teams call with a translation under each line. No bot joins the call, the extension listens to the browser tab itself.
English is not my first language, so I know the feeling the product is built on. You are on a work call, you follow most of it, you miss the one part that mattered, and you nod anyway. A lot of people do this and almost nobody says it out loud.
I can also talk about the money, because it nearly stopped the project. Speech recognition plus translation cost me about $2.55 for one hour of a call. Moving to one service that recognises and translates in the same stream brought it to about $0.154 per hour. The price of that was a few seconds more delay, not a better result.
I am open about the limits. Automatic captions get names wrong, they struggle with crosstalk and poor audio, they are no replacement for a human interpreter or for CART, and the tool only works when the meeting runs in a browser tab. There is no English to English mode yet.
I am happy to talk about any of this, including the parts that did not work. I record remotely and I am used to speaking English on calls.
