TAP-Software Requirements Specification (SRS)

🖇️ 1. Introduction

• Purpose: To build a transcription platform based on OpenAI's Whisper software to facilitate audio file transcriptions for a research team within a secure local network environment.
• Scope: The platform will consist of a web interface and local machine execution that handles audio file uploads, transcription, and speaker identification without requiring user login. This system will only be used within the research group.

🖋️ 2. General Description

• Product Functions:

  1. File upload via a simple web interface.
  2. Execution of transcription on the local machine.
  3. Email notification with transcription results.
  4. Identification of different speakers in the transcription.
  5. Short-term storage of transcriptions in a local database.

• User Characteristics: Researchers in residential care settings who require transcription of recorded group conversations for qualitative data analysis.

• Assumptions and Dependencies:

  • The platform will operate within a secure LAN.
  • Users will be limited to a small research team (about 3 concurrent users).

📑 3. System Features

Functional Requirements

Functional requirements describe the specific functions and features the system must provide, focusing on what the system is supposed to do.

• File Upload:

  • Support uploading audio files.
  • Support uploading video files. (optional)
  • The user can select a format for the output file. (Optional)
  • Drag-and-drop functionality for file uploads. (optional)
  • The user can input his email address to receive the transcription result.

• Transcription Processing:

  • Implementation of a queue system to handle multiple file transcriptions.
  • Use the up-to-date OpenAI Whisper model. (optional)
  • Ability to differentiate speakers labeled as "Speaker 1, 2, 3, etc."
  • Send transcription results to the user through email.
  • Send system error reports to users through email.
  • Support 'txt' output format.
  • Support 'docx' output format. (optional)
  • Support the Chinese language. (optional)

• File Management:

  • Save transcription result files in the local database.
  • Temporary storage of files with an automated deletion policy after 30 days.
  • Have an admin portal to see the history record.
  • The system can send a link through email to allow the user to enter the admin portal.
  • Download history transcription result files in the admin portal.

Non-Functional Requirements

Non-functional requirements describe how the system operates and the quality standards it must meet, focusing on how the system should perform.

• Security and Compliance:

  • Ensure secure access within the LAN.
  • No high-level security protocols specified, but data privacy is crucial due to sensitive research data.

• Interactivity and Accessibility:

  • Desktop-based web interface with minimal design.
  • No requirement for a mobile version.
  • Simple and user-friendly interface for uploading and receiving results.

• Performance Requirements:

  • The system should efficiently handle up to 3 concurrent (or in queue) transcriptions.
  • Expected to manage audio files up to 4 hours in length.
  • Improvement in transcription processing time, aiming for less than 4 hours for 1-hour audio files.
  • Optional use of additional open-source libraries to enhance speaker diarization capabilities.

• Security Requirements:

  • Ensuring data privacy for sensitive research information.
  • Implementation of network security measures within the LAN.