Misc software
TranscriTranslator
Voice dictation on a Stream Deck key, published on the Elgato Marketplace: transcribe, translate or rewrite.
The problem
Dictating text on a computer is still slow, full of steps, inflexible and short of quick triggers for translation and rewriting. I set out to solve it for use on the Stream Deck.
What was built
A Stream Deck plugin. Just press the key and speak, and the text appears in the active field, transcribed, translated into 28 languages or rewritten following the user’s instruction. It is published on the Elgato Marketplace, free, and open source under the MIT licence.
It runs on the user’s own OpenAI key, with no server in between: gpt-transcribe transcribes, and gpt-5.6-luna cleans up, translates or rewrites. Going by OpenAI’s price list, the estimate is half a US cent per spoken minute.
How it was built
TypeScript on the Stream Deck SDK, with ffmpeg recording the microphone. AI wrote much of the code; my part was testing, measuring and deciding. Every decision that must not be reverted is documented in the repository, along with the measurement behind it.
Gallery
Screens generated from the plugin’s real interface:
