Scriberr

Scriberr is an open-source, self-hostable AI application designed for fast audio transcription & summarization on the local.

It allows you to transcribe audio files directly on your own hardware by using OpenAI’s Whisper models and the Whisper.cpp inference engine. Furthermore, Scriberr integrates with OpenAI’s ChatGPT API to provide the possibility to create custom summaries of your transcripts.

Features

• Fast transcription with hardware acceleration: Supports a variety of platforms and optimizes performance based on available computing power.
• Customizable compute settings: Users can adjust the number of threads, cores, and select the model size for optimal performance.
• Local transcription: Audio files are processed directly on your device, giving you full control over data privacy.
• API integration: Scriberr exposes API endpoints for automating transcription processes or integrating with other tools.
• Summarization with ChatGPT: You can create summaries using your own prompts through the ChatGPT API.
• Mobile-friendly design: Scriberr is ready for use on mobile devices, enhancing accessibility.
• Ease of use: A straightforward, simple interface allows anyone to get started quickly.
• More features planned: Regular updates are expected to introduce new capabilities.
• Speaker diarization for labeling individual speakers.
• Audio recording capabilities for capturing new content directly in the app.
• Advanced file management with rename, delete, and organization options.
• Full-text fuzzy search for better navigation through transcripts.
• Tag-based organization for easier categorization of audio files.
• Real-time follow-along text playback to enhance review processes.
• More language support for transcription and summarization.

Use Cases

• Content Creators and Podcasters:
  If you’re creating audio or video content, Scriberr can help convert your recordings into transcripts quickly and accurately. The ability to summarize lengthy discussions using ChatGPT can further streamline the content creation process.
• Journalists and Researchers:
  For those conducting interviews or research, Scriberr can transcribe conversations or speeches, giving you written records without sending audio files to third-party servers. The summarization feature also aids in extracting key points from large volumes of data.
• Businesses and Corporate Users:
  Companies with large audio libraries or recorded meetings can use Scriberr to transcribe and summarize content for quick reviews or archival purposes. Its ability to run locally ensures sensitive data remains secure.
• Educators and Students:
  Lectures, presentations, or group discussions can be easily transcribed for future reference or study. Summaries can condense long transcripts into concise versions that capture the main points, perfect for note-taking or revision.
• Automation in Workflows:
  Developers can integrate Scriberr’s API into their existing workflows or automation pipelines. For instance, you can set up a system to transcribe customer service calls or virtual meetings in real-time, saving valuable time.

How It Works

Scriberr runs on Docker, which makes it easy to deploy across different environments. After setting up Docker and configuring your system, audio files are uploaded to the designated folders. The Whisper.cpp engine takes over the transcription process, using hardware acceleration based on the number of threads and cores you allocate.

Once the audio is transcribed, you can summarize the text using the ChatGPT API, where custom prompts allow for tailored results. The tool provides APIs that can be used to integrate Scriberr into larger systems or automation setups.

Pros

• Local processing ensures privacy by keeping sensitive audio data on your hardware.
• Customizable settings allow you to fine-tune performance based on your hardware capabilities.
• Versatile integration options through API endpoints, making it suitable for advanced workflows.
• Cross-platform support allows Scriberr to run on various devices with ease.
• Summarization feature adds extra value to the transcripts, especially for long or detailed content.

Cons

• Initial setup requires Docker, which might be a barrier for users unfamiliar with containerization tools.
• Beta version issues such as occasional bugs (e.g., initial database loading error) might require some troubleshooting.
• UI requires page refreshes in certain situations to reflect newly transcribed files or processed updates.

Installation Guide

To install Scriberr, you’ll need to run it via Docker. Here’s a simple docker-compose setup:

services:
  scriberr:
    image: ghcr.io/rishikanthc/scriberr:beta
    ports:
      - "3000:3000"
      - "8080:8080" # Optionally expose DB UI
      - "9243:9243" # Optionally expose JobQueue UI
    environment:
      - OPENAI_API_KEY=<your_openai_api_key>
      - [email protected]
      - POCKETBASE_ADMIN_PASSWORD=password
      - REDIS_HOST=127.0.0.1
      - REDIS_PORT=6379
      - SCRIBO_FILES=/scriberr
    volumes:
      - ./pb_data:/app/db
      - ./scriberr_files:/scriberr

Note: to create the necessary directories (audio and transcripts) inside the mapped volume before starting the service. On the first load, you may encounter a 500 error due to a missing database collection. Simply reload the page to resolve this issue.

Pricing

Scriberr is currently free and open-source. However, using the ChatGPT summarization feature requires an OpenAI API key, which has associated costs.

Known Bugs

• The first app load will fail due to a missing database, but a page refresh fixes it.
• Transcription results may not display immediately and require a manual page refresh.
• Automatic updates of processed files might be inconsistent and also require refreshing the UI.