VidBee
Downloads media, generates local transcripts with speaker labels, and runs custom AI analysis.
Extract searchable, timestamped transcripts with speaker detection and run AI analysis on your computer.
VidBee is an open-source desktop and web application that downloads media from over 1,000 sites or processes local files to generate searchable transcripts. It runs speech recognition directly on your computer and allows you to analyze audio with your chosen AI provider. Legal teams and investigators can use it to transcribe depositions, hearings, and interviews while keeping confidential media on local hardware.
What it is
VidBee transcribes audio and video files using local model families such as Whisper, SenseVoice, Parakeet, or Qwen3-ASR. It automatically identifies speakers, syncs transcript text to media playback timestamps, and provides built-in AI prompts for summaries, Q&A, and translations. You can connect it to local AI endpoints like Ollama or commercial APIs while maintaining control over your download queue and metadata.
Who it's for
This tool suits legal practitioners, investigators, and court reporters who need to convert audio or video evidence into searchable text without sending raw files to third-party services. It is less suited for organizations looking for automated live streaming transcription or teams that do not have the local hardware required to run speech recognition models.
Key features
Local speech recognition engine
Transcribe local files or downloads on your computer using Whisper, SenseVoice, Parakeet, or Qwen3-ASR without uploading media to external servers.
Speaker detection and timestamps
Identify speakers automatically, adjust speaker counts, search text, and jump directly to matching media timestamps.
Multi-provider AI prompt hub
Run summaries, translations, and custom queries against transcripts using local providers like Ollama or external API keys.
Automated RSS download queues
Subscribe to RSS feeds with keyword rules, custom directory templates, and automatic queuing to manage ongoing media releases.
Up and running in 3 steps
Our honest verdict
What's great
- Runs speech recognition locally to maintain privacy for confidential legal recordings.
- Connects to local LLMs via Ollama or LM Studio to keep text analysis off third-party clouds.
- Allows re-labeling speakers and editing speaker counts without re-processing speech recognition.
- Combines media downloading from 1,000+ supported sites with automated RSS channel monitoring.
Where it falls short
- Local speech recognition models require capable desktop hardware or dedicated GPU resources to run efficiently.
- Deploying the web interface and API server requires Docker or manual node monorepo management.
- Sending transcripts to external AI APIs requires extra caution to prevent accidental data disclosure.
Pricing & license
VidBee is free and open-source under the MIT licence. There are no paid hosted tiers provided in the project repository, so your only cost is the hardware infrastructure needed to run local speech and AI models or any external LLM API fees you choose to incur.
Open remote AI-training roles
767 open remote AI-training roles, updated November 14, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
Use it, or have us set it up
FAQ
Does VidBee upload my media files to third-party servers?
No. Speech recognition runs directly on your computer using local models like Whisper, keeping your audio and video files on your system.
Can I process transcripts without sending data to cloud AI services?
Yes. You can connect VidBee to local endpoints like Ollama or LM Studio so that both speech recognition and prompt processing remain local.
How does speaker identification work in VidBee?
VidBee can detect speakers automatically or let you set a fixed speaker count, allowing you to re-label speaker names without re-transcribing the text.
