Scriberr
Self-hosted AI audio transcription
Last verified:
What is Scriberr?
Scriberr is a self-hostable, offline audio transcription application that runs entirely on your local hardware. It uses the WhisperX engine with OpenAI's Whisper models for fast, accurate transcription with automatic language detection and support for all Whisper model sizes. The app is built with Go for the backend, Svelte for the frontend, and Python for AI transcription, packaged as a single binary for easy deployment.
Key features include speaker diarization for detecting and identifying multiple speakers with color-coded output, automatic summarization using OpenAI or Ollama endpoints, AI-powered chat with transcripts for taking notes, built-in audio recorder for instant transcription, YouTube video transcription capability, and multiple export formats (plaintext, JSON, SRT). The app also offers advanced parameter tuning for transcription and diarization models, audio playback synchronization with transcript highlights, and customizable summarization prompt templates.
Scriberr is designed for self-hosters, privacy-conscious users, researchers, content creators, podcasters, meeting note-takers, and anyone who needs to transcribe audio files locally without sending data to cloud services. It's particularly suitable for users who want full control over their data and prefer open-source solutions.
The application is currently in public beta (v1.0.0-beta1) and is completely free and open-source under the MIT license. It can be deployed via Docker or compiled from source, with storage using SQLite and audio files stored directly on disk.
Scriberr pricing
Pricing model: Freemium
Scriberr is completely free and open-source under the MIT license. There are no paid plans or subscription fees for the core application. The app itself has no cost. However, using the ChatGPT summarization feature and AI chat with transcripts requires an OpenAI API key, which has associated costs based on OpenAI's pricing. Alternatively, users can set up a self-hosted Ollama instance for fully free summarization and chat functionality. Installation options include a free Docker image (ghcr.io/rishikanthc/scriberr:v1.0.0-beta1) or building from source at no cost.
Scriberr pros
- Completely free and open-source under MIT license
- Fully offline transcription - no data sent to cloud
- Self-hostable with single binary deployment
- Fast transcription using WhisperX engine
- Supports all Whisper model sizes
- Automatic language detection
- Speaker diarization with color-coded speakers
- Built-in audio recorder for instant transcription
- YouTube video transcription support
- Multiple export formats (TXT, JSON, SRT)
- AI chat with transcripts via OpenAI/Ollama
- Automatic summarization with customizable prompts
- Audio playback sync with transcript highlights
- Advanced parameter tuning for models
- Clean and intuitive UI with Svelte frontend
- Multiple chat sessions per transcript
- Save and reuse prompt templates
- CUDA support for GPU acceleration
- Toast notifications for instant feedback
- Right-click context menu for quick actions
Scriberr cons
- GPU support requires compiling custom Docker image
- Currently in beta version (not stable release)
- Chat/summarization requires OpenAI API key (paid)
- Requires self-hosting technical knowledge
- Need to install dependencies for building from source (npm, uv, ffmpeg, yt-dlp, go)
- No official progress bar for transcription status
- Diarization component provides no progress updates
- No pre-built CUDA Docker image available yet
- Session key must be generated manually for production
Frequently asked questions about Scriberr
What is Scriberr?
Scriberr is a self-hostable, offline audio transcription application that uses the WhisperX engine with OpenAI's Whisper models for fast transcription. It features automatic language detection, speaker diarization, support for all Whisper model sizes, and runs entirely on your local hardware without sending data to cloud services.
Is Scriberr free to use?
Yes, Scriberr is completely free and open-source under the MIT license. There are no subscription fees or paid plans for the core application. However, using the ChatGPT summarization feature requires an OpenAI API key which has associated costs, or you can set up a self-hosted Ollama instance for free summarization.
How do I install Scriberr?
Scriberr can be installed in two ways: using the provided Docker image with docker-compose (image: ghcr.io/rishikanthc/scriberr:v1.0.0-beta1), or by compiling from source. For source compilation, you need npm, uv, ffmpeg, yt-dlp, and Go installed. The app runs on port 8080 and uses SQLite for database management.
Does Scriberr work offline?
Yes, Scriberr operates completely offline for transcription. All transcription processes are conducted on your device using the WhisperX engine, ensuring no audio information is transmitted to any cloud services. However, chat and summarization features require either OpenAI endpoints or a self-hosted Ollama instance.
What audio formats does Scriberr support?
Scriberr supports audio files that Whisper can process. The app uses ffmpeg for audio processing, which supports most common audio formats. Files are stored directly on disk, and the app can also transcribe YouTube videos directly using yt-dlp integration.
Can Scriberr identify multiple speakers?
Yes, Scriberr includes speaker diarization that detects and identifies different speakers in your audio. The feature uses advanced algorithms and displays speakers with color-coding in the transcript. Enhanced diarization has been substantially improved in the beta version.
What export formats are available?
Scriberr supports three export formats: plaintext (TXT), JSON, and SRT (subtitle format). These can be downloaded directly from the app for use in other applications or for sharing.
Can I chat with my transcript?
Yes, Scriberr features interactive AI chat with your transcripts, powered by OpenAI or Ollama endpoints. You can ask questions about the transcript, get summaries, and take notes. The app supports multiple chat sessions for each transcript, allowing you to organize different conversations separately.
Does Scriberr support YouTube transcription?
Yes, Scriberr can transcribe YouTube videos directly. The app uses yt-dlp to download and process YouTube audio, making it easy to transcribe video content without manually downloading files first.
What programming languages is Scriberr built with?
Scriberr uses Go for the backend, Svelte for the frontend, and Python for AI transcription. The frontend is compiled to a static SPA (plain HTML and JavaScript) which is embedded into the Go backend binary, creating a single executable binary for easy deployment.