Transcription that knows your subject

Why transcripts usually go wrong

Real rooms are hard to transcribe: the speaker is far from your microphone, people nearby talk, and the words that matter most are the ones a speech recogniser has never seen, such as a client’s name, a product code or the acronyms of your field. Audiken deals with this twice. Before recording, it builds a list of up to fifty names, acronyms and terms from your files and tells the speech recogniser to listen for them. After recording, it reads the whole transcript against those files and corrects what still came out wrong.

What the review changes

The review fixes misheard words where your files make the right word clear, turns numbers and formulas that were spoken aloud into written ones, removes conversations that were not part of the session, and marks anything it could not make out instead of guessing. It then gives the session a title, a topic and a short summary. Every change is listed, and the untouched original is kept beside the reviewed version.

Nothing lost when the connection drops

Wi-Fi fails at the worst times. If the connection drops, Audiken keeps recording and holds the sound in your browser, for up to half an hour of lost connection. When it returns, it catches up at double speed, stamps each line with the time it was actually spoken, and tells you how far behind it is until it is level again.

What you can do with it afterwards

A transcript is the raw material. Once a session is in your space, you can ask questions about it and get a summary, action items or a follow-up with sources; build flashcards from it; and get practice tests that weigh what was emphasised.