@rhs Sure! We try to automate away anything that machines can do well, and humans do the rest. Transcription is actually made of smaller problems — detecting speech, labeling speakers, audio to text, etc. We use a bunch of tools for each task, some of it is custom on top of open source (Kaldi - http://kaldi-asr.org/ for example), but we also use other services/APIs, and of course the humans come in at the end. We want to abstract everything away and just find the best option out there, so that we can provide a great user experience.
Jotengine
Wefunder
Crisp
Rainforest QA
Jotengine
Story Replay