Offline-first forensic translation and transcription tool. For law enforcement and intelligence analysts.
AUGUR surfaces foreign-language content inside digital evidence — text, audio, video, image, PDF, subtitle — and translates it into the examiner's working language. No evidence ever leaves your machine.
- Detects the language of evidence in 16+ (whichlang) or 176 (fastText) languages.
- Transcribes audio / video with Whisper (Metal-accelerated on Apple Silicon), with per-segment timestamps + optional speaker diarization.
- Translates with NLLB-200-distilled-600M (200 languages), with a faster ctranslate2 backend when available.
- OCRs images with Tesseract; extracts PDF text layers; parses SRT/VTT subtitles; geolocates IPs against MaxMind GeoLite2; converts forensic timestamps (Unix / Apple / Windows / WebKit / HFS+).
- Scans translated + original content with YARA rules.
- Bundles results into evidence packages with SHA-256 manifests, agency-branded HTML reports, and a chain-of-custody header.
# Translate a single audio file
augur translate --input recording.mp3 --target en
# Walk an evidence directory and emit an HTML report
augur batch --input /evidence --target en \
--format html --output report.html --threads 4
# Bundle the run into a shareable evidence package
augur package --input /evidence --output case-001.zipgit clone <augur repo> augur && cd augur
cargo build --release
sudo install target/release/augur /usr/local/bin/augur
augur self-test --full # downloads models on first useFor air-gapped workstations: see
docs/AIRGAP_INSTALL.md.
Every translation AUGUR produces is labeled machine-generated. The advisory fires on every output surface — console, batch JSON / CSV / HTML, evidence manifest, Strata plugin artifact — and is not suppressible by any flag.
For legal proceedings, verify all translations with a certified human translator.
docs/USER_MANUAL.md— full feature reference for examiners.docs/QUICK_REFERENCE.md— one-page command cheat sheet.docs/DEPLOYMENT.md— workstation setup + casework workflow.docs/AIRGAP_INSTALL.md— offline install for classified environments.docs/STRATA_INTEGRATION.md— building AUGUR as a Strata plugin.docs/LANGUAGE_LIMITATIONS.md— known classifier ambiguities (Pashto/Persian, Hindi/Urdu, short-text reliability) and how AUGUR surfaces them.CLAUDE.md— engineering invariants + sprint decisions for developers.
Proprietary — Wolfmark Systems.