CLI Commands

Augent provides a full CLI for terminal-based workflows.

augent search

Search audio files for keywords with timestamped results.

augent search <audio> "keyword1,keyword2"

Flag	Description
--model, -m	Model size (default: tiny)
--full, -f	Include full transcription in output
--format	Output format: json, csv, srt, vtt, markdown
--output, -o	Write output to file
--export-clips DIR	Extract audio clips around matches
--clip-padding	Seconds before/after each clip (default: 15)
--workers, -w	Parallel workers for batch processing
--stream, -s	Stream transcription progress
--no-cache	Disable caching
--quiet, -q	Suppress progress messages

---

augent transcribe

Full transcription without keyword search.

augent transcribe <audio>

Flag	Description
--model, -m	Model size (default: tiny)
--format	Output format: json, srt, vtt
--output, -o	Write output to file
--stream, -s	Stream progress
--no-cache	Disable caching

---

augent proximity

Find two keywords appearing near each other.

augent proximity <audio> "keyword1" "keyword2"

Flag	Description
--distance, -d	Max words between keywords (default: 30)
--model, -m	Model size
--format	Output format: json, csv, markdown
--output, -o	Write output to file

---

augent memory

Manage transcription memory.

augent memory stats          # View memory statistics
augent memory list           # List stored transcriptions by title
augent memory clear          # Clear all stored transcriptions
augent memory clear-models   # Clear loaded models from memory

---

audio-downloader

Download audio from any video URL. 16 parallel connections, audio-only extraction, no conversion.

audio-downloader "URL"
audio-downloader "URL1" "URL2" "URL3"

• aria2c multi-connection downloads (16 parallel connections)
• Concurrent fragment downloading (4 fragments)
• No video download, audio extraction only
• No format conversion, native audio format

Supports YouTube, Vimeo, SoundCloud, Twitter/X, TikTok, Twitch, Reddit, and 1000+ more via yt-dlp.

---

Examples

# Download and transcribe
audio-downloader "https://youtube.com/watch?v=xxx"
augent transcribe ~/Downloads/video.webm

# Search with CSV export
augent search podcast.mp3 "open source,self-hosted" --format csv --output results.csv

# Batch process a folder
augent search "*.mp3" "keyword" --workers 4

# Export subtitles
augent transcribe lecture.mp3 --format srt --output subtitles.srt

# Extract audio clips around matches
augent search interview.mp3 "important" --export-clips ./clips --clip-padding 15

# Proximity search
augent proximity podcast.mp3 "problem" "solution" --distance 30