# Deepgram

> Deepgram CLI Skills Guide

- Skill: `demerzels-lab/deepgram` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add demerzels-lab/deepgram`
- Raw SKILL.md: https://api.skillmd.com/api/skills/demerzels-lab/deepgram/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: Demerzels-lab (https://skillmd.com/u/demerzels-lab)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/demerzels-lab/deepgram

---

# Deepgram CLI Skills Guide

## Tool
**@deepgram/cli** — command-line interface for Deepgram speech-to-text.

---

## Install
```bash
npm install -g @deepgram/cli
```

---

## Auth

```bash
deepgram login
```

Uses your Deepgram API key (stored locally).

---

## Core Skill: Speech → Text

### Transcribe a Local Audio File

```bash
deepgram listen prerecorded audio.wav
```

### Transcribe with Options

```bash
deepgram listen prerecorded audio.wav \
  --model nova-2 \
  --language en \
  --punctuate \
  --diarize
```

---

## Core Skill: Read / Reach Content

### From URL (remote audio)

```bash
deepgram listen prerecorded https://example.com/audio.mp3
```

### From STDIN (pipes)

```bash
cat audio.wav | deepgram listen prerecorded -
```

### From Microphone (live)

```bash
deepgram listen microphone
```

Stop with `Ctrl+C`. Congrats, you just dictated reality.

---

## Output Handling

### Save Transcript

```bash
deepgram listen prerecorded audio.wav > transcript.json
```

### Plain Text Output

```bash
deepgram listen prerecorded audio.wav --format text
```

---

## Useful Flags (Memorize These)

* `--model` – `nova-2`, `general`, etc.
* `--language` – `en`, `tr`, `de`, …
* `--punctuate` – adds punctuation
* `--diarize` – speaker separation
* `--format` – `json`, `text`, `srt`, `vtt`

---

## Typical Workflow

1. Reach content (file / URL / mic)
2. Run `deepgram listen`
3. Capture output (JSON or text)
4. Post-process (search, summarize, subtitle)

---

## Skill Summary

* CLI-based speech-to-text
* Local, remote, and live audio
* Scriptable, pipe-friendly
* Fast, accurate, no UI nonsense

Deepgram CLI: because keyboards are overrated.