Allow user to flag incorrect Speech to Text transcription

[demilolu]

Background on the problem the feature will solve/improved user experience

Speech to Text transcription errors may occur in recordings that may lead in erroneous inconsistencies in our AI model.

Describe the solution you'd like

Provide a means of user/crowdsource flagging of incorrect phrases in speech to text transcription so that police-eye witness inconsistencies are not attributed to machine error.

Proposed user flow:

• Review a transcription
• Find a word that doesn't make sense
• listen to the recording at that point
• flag that phrase/time stamp
• flagging is added to blockchain as well
• downstream third party/admin is able to review that flag and evaluate if any found police-eye witness inconsistencies were related with that flagged

Concerns: this might also be abused so we'd have to find a way to mitigate against manipulating that.

Tasks

• investigate what percent of inconsistencies can be due to speech to text errors

Acceptance Criteria

• As a user, I'm able to read a transcribed report
• As a user, I'm able to flag individual words that might be transcribed incorrectly at the appropriate timestamp in the recording and the transcript
• As a member of the public, i can see all the words that have been flagged openly on the blockchain
• As an administrator, I'm able to investigate if inconsistencies found by AI are due to speech-to-text mistakes.

Out of Scope

• Further crowdsourcing to "upvote" flagged words

[github-actions[bot]]

:wave: Hi! This issue has been marked stale due to inactivity. If no further activity occurs, it will automatically be closed.