README.md

metadata

title: Knowledge Transcriber
emoji: 🦀
colorFrom: red
colorTo: green
sdk: gradio
sdk_version: 4.26.0
app_file: app.py
pinned: false

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

Knowledge Transcriber - Self Service Guide

Introduction

The Knowledge Transcriber project automates the generation of transcripts from video recordings, tailored for self-service use. It leverages a Python script to process videos stored in a publicly accessible Google Drive folder, generating outputs that include high-level summaries, topic-specific summaries, glossaries, and more. The project is designed to facilitate departments in managing and paying for their translations independently. For more information and source code of the Python script, visit GitHub repository.

Key Features

• Automated Transcription: Converts video recordings into detailed transcripts.
• Self-Service Interface: Utilizes Gradio for a simple, user-friendly interface.
• Google Drive Integration: Supports linking to video recordings and storing outputs on Google Drive.
• Flexible API Key Input: Allows different departments to use their own OpenAI API keys.

Deployment

The project is deployed on HuggingFace Spaces, available at https://huggingface.co/spaces/skyvera/kt-transcribe. Any changes made to the code should be uploaded to https://huggingface.co/spaces/skyvera/kt-transcribe/tree/main. Once the modified file is uploaded, the app will automatically rebuild.

Usage

1. Prepare Your Environment: Prerequisites such as Python 3.11 and ffmpeg are automatically managed by Hugging Face Spaces through the requirements.txt and packages.txt files.
2. Set Up Google Drive: Make sure the video recording link on Google Drive is publicly accessible and has public access permissions for writing.
3. Launch the Application: Run app.py to start the Gradio interface.
4. Input Details: Enter your OpenAI API key and the Google Drive link to the video recording
5. Process and Access Outputs: The application will process the video and store the outputs in the shared Google Drive folder.

Important Notes

• Security: Do not upload .env files or any sensitive information to the HuggingFace repository.
• API Key: The OpenAI API key is required for transcription services. Ensure it is kept secure and not exposed in the code.
• Google Drive Permissions: Ensure the video recording folder link is set to public access to avoid permission issues.

Example Code Changes for Deployment

To adapt the script for self-service and deployment, modifications are centered around the Gradio interface setup in app.py and ensuring Google Drive integration works seamlessly. The core logic for processing videos remains unchanged.

Gradio Interface Setup

demo = gr.Interface(
    process_inputs,
    [
        gr.Textbox(label="OpenAI Key", info="Enter your OpenAI API Key."),
        gr.Textbox(label="Drive Folder Link", info="Enter your Drive Folder Link for video recordings."),
        gr.Textbox(label="Output Folder Link", info="Enter your Drive Folder Link for storing outputs."),
    ],
    gr.Textbox(label="Result"),
    theme=gr.themes.Base()
)
demo.launch()

Google Drive Integration

Ensure the KnowledgeTranscriber class in kt_transcript.py correctly handles Google Drive file listing, downloading, and uploading.

Conclusion

The Knowledge Transcriber project simplifies the process of generating transcripts from video recordings, making it accessible for various departments to manage their transcription needs independently. With its deployment on HuggingFace Spaces, users can easily access and use the application through a web interface.