Meeting Automation
Budget: $250 – $750 USD
Meeting Audio Transciption Automation
Scope of Work
The scope of work for this project is to implement meeting automation for transcribing meetings from invites. The integrations should be implemented for Google Meet, Teams, and Zoom)
- The automation is to be deployed to meetings, invited by an email alias (ex. something @ website. io)
- Users will invite an email alias for the automation ahead of a scheduled meeting through the meeting platforms scheduling feature
- The automation is to be deployed to meeting url’s based on a user submission on the front end
- The automation is to record audio of meeting for the length of the meeting and until all participants have left
- The automation is to collect meeting data:
- Emails and names of attendants and those invited to email
- Length of meeting
- Time of Meeting
- Any additional information for diarization
The automation will generate a transcript for each meeting, as well as a headline, summary, keywords, action items, follow-ups, decisions, and ambiguity.
- Using the Whispr model deployed on Modal, the the audio will be fed through the model to create the transcript and whatever additional technical needs to create a diarization of the speakers (who said what and when)
- Based on the transcript, the text will be fed through OpenAI models to generate a series of items structured in a json.
- Keywords (Curie or Babbage)
- Headline of the meeting (Curie or Babbage)
- Summary of the meeting (Curie or Babbage)
- Action Items (DaVinci)
- Decisions (DaVinci)
- Follow-ups (DaVinci)
We will be using Modal for implementing the automation and Supabase for storage, database, and authentication purposes.
The API should be setup for streamlined integration with the front end (the front end is through the weweb platform)
Scope of Work
The scope of work for this project is to implement meeting automation for transcribing meetings from invites. The integrations should be implemented for Google Meet, Teams, and Zoom)
- The automation is to be deployed to meetings, invited by an email alias (ex. something @ website. io)
- Users will invite an email alias for the automation ahead of a scheduled meeting through the meeting platforms scheduling feature
- The automation is to be deployed to meeting url’s based on a user submission on the front end
- The automation is to record audio of meeting for the length of the meeting and until all participants have left
- The automation is to collect meeting data:
- Emails and names of attendants and those invited to email
- Length of meeting
- Time of Meeting
- Any additional information for diarization
The automation will generate a transcript for each meeting, as well as a headline, summary, keywords, action items, follow-ups, decisions, and ambiguity.
- Using the Whispr model deployed on Modal, the the audio will be fed through the model to create the transcript and whatever additional technical needs to create a diarization of the speakers (who said what and when)
- Based on the transcript, the text will be fed through OpenAI models to generate a series of items structured in a json.
- Keywords (Curie or Babbage)
- Headline of the meeting (Curie or Babbage)
- Summary of the meeting (Curie or Babbage)
- Action Items (DaVinci)
- Decisions (DaVinci)
- Follow-ups (DaVinci)
We will be using Modal for implementing the automation and Supabase for storage, database, and authentication purposes.
The API should be setup for streamlined integration with the front end (the front end is through the weweb platform)