Request for Information (RFI) for Transcribing Software

Background Case files processed by Criminal Justice include transcripts from suspect interviews and victim interview transcripts.

These transcripts are verbatim records of interviews (audio and video) uploaded onto the Police digital evidence system.

A team of transcribers currently retrieve these interviews and transcribe them onto the MG15 Record of Interview Form.

The transcripts are required by the Crown Prosecution Service (CPS), defence and HMCTS in specific cases.

As the type of requests are expanding and the length and number of interviews is increasing, there is now a need to look at potential automated solutions in the form of transcribing technology to cope with the increasing demand and deliver the required transcripts in an efficient and cost-effective way.

Requirements o Must be able to transcribe video and audio from different sources e.g.

DIR, Body Worn Video, MP4, Achieving Best Evidence (ABE) etc o Must be able to transcribe quickly (currently it takes 60 minutes to transcribe 10 minutes of audio, and it is anticipated that an software solution would be able to do this quicker and more effectively/efficiently) o Must be able to transcribe different accepts and tones of voices. o Must have a high accuracy rate o Must be able to separate the speakers and identify speaker 1,2,3 etc. o Must be able to transcribe onto a template (MG15) o Must be able to highlight the areas where the AI/Software is not certain its accurate to enable an individual to check and verify those particular areas o Must be able to connect to NICE where our interviews are stored.