Voice to Text

Push-to-talk voice transcription plugin for Obsidian.
Hold a hotkey → speak → release → text is inserted at the cursor.

Providers: Deepgram (Nova 2) · Groq (Whisper)
Author: SATOSprod
License: Proprietary — see LICENSE

---

Features

• Push-to-talk — hold a key combination to record, release to transcribe
• Two providers — Deepgram Nova 2 or Groq Whisper, switchable in settings
• Static model & language lists — dropdown selectors, no free-text input
• Interactive hotkey capture — click the field, press your combo, done; keys are displayed as icons
• File logging — activity logs written to .log files inside the vault (one file per day)
• Audio saving — optionally save each recording as a WAV file in the vault
• No emoji — SVG icons only in the status bar and settings UI
• Minimal flat design — no shadows, no hover animations

---

Requirements

• Obsidian 0.15.0 or later (desktop only)
• A Deepgram or Groq API key (free tiers available)
• Node.js 16+ and npm (for building from source)

---

Installation

From source (recommended)

# 1. Clone the repository
git clone https://github.com/SATOSprod/voice-to-text.git
cd voice-to-text

# 2. Install dependencies
npm install

# 3. Build
npm run build
# Produces: main.js

Then copy the plugin folder into your vault:

<your-vault>/.obsidian/plugins/voice-to-text/
├── main.js          ← compiled output
├── manifest.json
├── styles.css

Open Obsidian → Settings → Community plugins → Installed plugins and enable Voice to Text.

Development mode (auto-rebuild on save)

npm run dev

---

Configuration

Open Settings → Voice to Text.

Setting	Default	Description
Provider	Deepgram	Switch between Deepgram and Groq
API key	—	Secret key for the chosen provider
Model	nova-2 / whisper-large-v3	Select from a fixed list per provider
Language	Auto	Dropdown: auto-detect or a specific language code
Hotkey	Meta+Alt	Click the field to capture interactively
Enable logging	Off	Write activity logs to the vault
Log folder	voice-to-text-logs	Vault-relative path; created automatically
Save recordings	Off	Keep a WAV file of each recording
Recordings folder	voice-recordings	Vault-relative path; created automatically

Getting an API key

Deepgram

1. Go to console.deepgram.com
2. Create a free account (includes $200 credit)
3. API Keys → Create a key → copy and paste into plugin settings

Groq

1. Go to console.groq.com
2. Create a free account
3. API Keys → Create API key → copy and paste into plugin settings

---

Usage

1. Open any note in Obsidian
2. Place the cursor where you want the text inserted
3. Hold your configured hotkey (default: Meta+Alt = Win+Alt / Cmd+Alt)
4. Speak — a notice appears confirming recording is active
5. Release the keys — transcription starts automatically
6. The transcribed text is inserted at the cursor position

If no editor is active, the transcription is copied to the clipboard instead.

---

Supported Languages

Code	Language	Code	Language
auto	Auto-detect	ko	Korean
ru	Russian	nl	Dutch
en	English	pl	Polish
de	German	tr	Turkish
fr	French	ar	Arabic
es	Spanish	uk	Ukrainian
it	Italian	zh	Chinese
pt	Portuguese	ja	Japanese

---

Supported Models

Deepgram

Model	Description
nova-2	Best accuracy, recommended
nova-2-general	General purpose
nova-2-meeting	Optimised for meetings
nova-2-phonecall	Optimised for phone audio
nova	Previous generation
enhanced	Legacy enhanced
base	Legacy base

Groq (Whisper)

Model	Description
whisper-large-v3	Best accuracy
whisper-large-v3-turbo	Faster, slightly lower accuracy
distil-whisper-large-v3-en	English-only, fastest

---

File Structure

voice-to-text/
├── main.ts           ← TypeScript source (single file)
├── main.js           ← compiled output (gitignored, built locally)
├── styles.css        ← plugin styles
├── manifest.json     ← Obsidian plugin manifest
├── package.json
├── tsconfig.json
├── esbuild.config.mjs
├── versions.json
├── .gitignore
├── LICENSE
└── README.md

---

License

This project is released under a proprietary license.
Copying source code into other projects is not permitted.
See LICENSE for full terms.

© 2026 SATOSprod