Transcribing spoken dialogues


Beskrivning

We need to transcribe a set of audio recordings. To this end, we have applied automated speech recognition to our dataset, but these automatic transcriptions are not perfect. Now we need to correct them.


Your task will be to listen to the audio recordings together with their transcriptions and correct any mistakes you spot.


The dataset we need to transcribe has approximately 10 hours of recordings of two persons in a dialogue (one microphone per person, so approximately 20 hours total).


We are looking to recruit 3 or 4 people, each of which is expected to spend approximately 50 hours on correcting their part.


Requirements: Proficient in English, comfortable with using a PC, Windows OS


We need to transcribe a set of audio recordings. To this end, we have applied automated speech recognition to our dataset, but these automatic transcriptions are not perfect. Now we need to correct them.


Your task will be to listen to the audio recordings together with their transcriptions and correct any mistakes you spot.


The dataset we need to transcribe has approximately 10 hours of recordings of two persons in a dialogue (one microphone per person, so approximately 20 hours total).


We are looking to recruit 3 or 4 people, each of which is expected to spend approximately 50 hours on correcting their part.


Requirements: Proficient in English, comfortable with using a PC, Windows OS


Plats

AB Transkibera


Kontakt

transkribera@mera.se