DVR Transcriber

Status: Work in Progress

A CLI utility and workspace template for importing, organizing, and transcribing audio files from Digital Voice Recorders (DVRs).

The Problem

Digital voice recorders produce flat directories of MP3 files with minimal metadata. Managing, organizing, and transcribing these recordings is tedious and time-consuming.

The Solution

A streamlined workflow that:

1. Imports recordings from a mounted DVR
2. Organizes files into date-based folders (DDMM format)
3. Classifies recordings using AI (via short audio samples)
4. Transcribes selected recordings with speaker identification
5. Exports transcripts and summaries to cloud storage

Planned Features

Import & Organization

• CLI for syncing from DVR mount point
• Move (default) or copy modes
• Auto-cleanup of accidental recordings (<10 seconds)
• Date-based folder organization

AI-Powered Classification

• Extract 30-second samples from recordings
• Send to Gemini for title and summary generation
• Avoid processing full multi-hour recordings just for metadata

Transcription

• Selective full transcription via Gemini API
• Voice sample-based speaker diarization
• Accurate speaker labels without manual intervention

Evidence Mode (Optional)

For recordings that may serve as legal evidence:

• No file modifications or renaming
• SHA256 checksum calculation
• Cloud backup (Google Drive, S3, WORM storage)

Use Cases

• Meeting notes and discussions
• Personal voice memos
• Research interviews
• Property viewing notes
• Any scenario where voice capture beats typing

Requirements (Planned)

• Linux
• Python 3.10+
• Gemini API key
• Mounted DVR (USB mass storage)

Installation

Coming soon

Usage

Coming soon

License

TBD