Logo
Documentation

BeltoVox: The Ultimate User Guide

Everything you need to know to turn your voice into perfectly formatted, polished text in any application.

Welcome to BeltoVox, a professional-grade Windows dictation tool that uses state-of-the-art AI to turn your voice into perfectly formatted, polished text in any application.

1

Getting Started: The API Keys

BeltoVox is a "Bring Your Own Key" (BYOK) application. This ensures your data stays private and you only pay for what you use directly to the AI providers.

Option A: Groq

Recommended - Fast & Free/Cheap

Groq is the fastest way to use BeltoVox. Transcriptions are nearly instantaneous.

  • Sign Up: Go to Groq Console.
  • Generate Key: Navigate to API Keys in the sidebar. Click Create API Key.
  • Copy: Save the key (starts with gsk_).
  • Pricing: Groq currently offers a generous Free Tier. For paid usage, it is typically priced by "tokens," but is extremely cost-effective.

Option B: OpenAI

Industry Standard
  • Sign Up: Go to OpenAI Platform.
  • Add Credits: OpenAI requires a prepaid balance (minimum $5). Go to Settings > Billing.
  • Generate Key: Go to API Keys and click Create new secret key.
  • Copy: Save the key (starts with sk-).
  • Pricing:
    • Whisper (Speech): ~$0.006 per minute of audio.
    • GPT-4o-mini (Refining): Fractions of a cent per request.
2

The Dashboard: A Deep Dive

Right-click the tray icon and select Show Dashboard to access these tabs.

Tab 1: API Settings

  • Provider Selection: Choose between OpenAI or Groq.
  • Credentials: Paste your encrypted keys here.
💡
You can save keys for both and switch between them anytime.

Tab 2: Triggers & Output

  • Global Keyboard Trigger: Set your favorite shortcut (Default: Ctrl + Space). This works even if BeltoVox is hidden.
  • Mouse Triggers:
    • Button: Map dictation to Side Button 1, 2, or Middle Click.
    • Mode: Use Push-to-Talk (hold to speak) or Toggle (tap to start/stop).
  • Status Overlay: Enable a floating circle that stays on top and shows if you are recording.
  • Delivery Mode:
    • Direct Insert: Types the text instantly at your cursor.
    • History Only: Only saves the text to the dashboard for later manual copying.

Tab 3: AI & Language

  • Input Language: Choose your native language or use Auto-detect. We support English, Hungarian, German, Thai, and more.
  • AI Processing Modes:
    • Standard: Literal transcription.
    • Refiner: Cleans grammar and removes "filler" words.
    • Professional: Rewrites your speech into a formal business tone.
    • Friendly: Perfect for quick, warm chat messages.
  • Translate to English: Automatically translates any foreign speech into English text.
  • Smart Formatting: Handles complex punctuation and capitalization automatically.
  • Word Corrections: Click Manage word list… to teach BeltoVox how to spell names, brands and jargon it tends to mishear (e.g. your company name). Your list stays on your device—nothing is ever uploaded. See the Pro Tips section below for how to use it.

Tab 4: Audio

  • Microphone: Select specific hardware if you have multiple mics.
  • Sample Rate: High quality (44.1kHz) or Optimized (16kHz). 16kHz is recommended for faster AI processing.
  • Feedback: Adjust the Beep Volume or disable sounds entirely.

Tab 6: Usage & Stats

  • Recording Time: Total time spent speaking.
  • API Total Requests: How many times the AI was called.
  • Est. Spend: A calculated estimate of your cost (using standard OpenAI/Groq rates) to help you track your budget.
3

Visual Indicators (Status Overlay)

The floating circle changes color to keep you informed without looking at the app:

Idle

Ready to listen.

Recording

Capturing your voice.

Paused

Recording is on hold.

Processing

AI is typing or refining.

4

Quick Control & System Tray

Left-click Tray Icon

Opens a compact "Quick Control" window. This is perfect for when you want a minimal interface while working.

Right-click Tray Icon

  • Quickly switch input languages.
  • Access Settings.
  • Open "Launch on Startup" options.
5

Pro Tips for Better Accuracy

Fix Mistranscribed Names

Notice a name, brand or place BeltoVox keeps getting wrong? Open AI & Language → Manage word list… and add a row. On the right, type the correct spelling (e.g. a brand or name like Anthropic) — that alone is usually enough, because BeltoVox automatically catches new mishears that sound like your word (Anthropik, Antropic…) without you listing them. Full details are in the Word Corrections section below.

Short Bursts

AI works best with clips between 5 seconds and 1 minute.

Clear Environment

Background noise can confuse the "Refiner" mode.

Administrator Rights

If your hotkey doesn't work in certain apps (like Task Manager), run BeltoVox as Administrator.

6

Word Corrections: Your Personal Dictionary

Speech-to-text sometimes mishears the words that matter most to you — a person's name, your company, a product or a place. Word Corrections lets you teach BeltoVox the right spelling once, and it fixes that word from then on. Everything here stays on your computer — your list is never uploaded.

Adding a correction

Open AI & Language → Manage word list… and add a row:

  • Right box — the correct spelling (required). For example, a brand or name like Anthropic. In most cases this is all you need to type.
  • Left box — wrong forms (optional). If a particular misspelling keeps appearing, type it here. You can list several, separated by commas — e.g. Anthropik, Antropic.
💡
Build the list up over time: whenever you spot a word BeltoVox keeps getting wrong, just add it. You can also Import… a list from a file or Export… yours to another computer.

Automatic catch of similar-sounding mistakes

You don't have to predict every wrong spelling. BeltoVox also catches new mishears that simply sound like one of your correct words. So if you added Anthropic, it will also fix Anthropik or Antropic even if you never listed them. Ordinary everyday words that happen to sound similar are deliberately left alone, so your normal sentences are never changed. This runs entirely on your device and adds no noticeable delay.

💡
If you ever see a real, everyday word getting changed by mistake, untick Auto-catch similar-sounding mishears in the same window. BeltoVox will then only correct the exact forms you listed.

The language word lists (you usually never touch these)

To decide which words count as "ordinary," BeltoVox keeps a small list of common everyday words for each language. These live in plain text files on your computer, and you almost never need to open them — they simply work in the background. They're editable only for the rare case where you want to fine-tune things yourself.

If you ever do want to look or make a change:

  • Open this folder on your PC: %APPDATA%\BeltoVox\word_guards (paste that into the address bar of a File Explorer window).
  • Each language has its own file, named common_words_<code>.txt — e.g. common_words_en.txt for English or common_words_hu.txt for Hungarian. The <code> is the two-letter language code (it = Italian, es = Spanish, and so on).
  • Each file is simply one word per line. Add a word to protect it from being auto-changed; remove one if you would rather BeltoVox treated it as a name. Lines starting with # are notes and are ignored. Save the file and restart BeltoVox to load your changes.
💡
Your edits are safe across updates — BeltoVox never overwrites a word list you've changed. English and Hungarian come especially well filled in out of the box.
Ready to Start? Get BeltoVox Now