hyperaudio / stt-benchmarking

a place to discuss and create an independent speech-to-text benchmarking tool
5 stars 1 forks source link

R12 - Guidelines for the reference text file #14

Open EyalLavi opened 5 years ago

EyalLavi commented 5 years ago

This is a documentation requirements. We will be converting the reference transcript into our internal format so inevitably we will have to decide how we deal with things like white spaces and hyphens (one word or two?). Our decision might not be obvious to the user, so we should spell it out.