Technical Documentation

Getting Started & Installation

Productive Voice is delivered as a signed Windows installer. The app runs as a lightweight background process and injects text directly into the active window via the Win32 API, with no visible browser chrome.

# Verify release executable integrity before launching
CertUtil -hashfile "Productive Voice Setup 0.3.1.exe" SHA256

To run the installer, launch the `Productive Voice Setup 0.3.1.exe` binary. The installer sets up files inside your user app directories and registers the global hotkey triggers within the OS registry so it wakes up instantly on startup.

Secure API Key Configurations

Productive Voice routes your audio directly to your provider's servers. Rather than passing your recordings through intermediate telemetry servers, it connects straight from your machine to the API endpoint under your own credentials.

Once keys are submitted via the settings capsule panel, they are encrypted locally via native AES-256-GCM methods derived from your active Windows account login keys. They can never be scraped by sandboxed software or background applications.

Your First Dictation Trigger

Transcription begins as you speak. You do not need to highlight a text field or toggle any widgets first. Position your cursor inside any text input, document, editor, or spreadsheet, then trigger the hotkey.

Press the default physical keyboard hotkey to wake the capsule dock:

Ctrl + Shift + Space

A floating pink pill will animate into view in the top margin of your workspace, marked with a pulsing recording dot. Speak your thought. Press the key bind again to terminate capture. Productive Voice streams audio to the provider, applies your custom dictionary substitutions, runs the formatting prompt, and pastes the clean text at your cursor.

Native Windows File Directories

Every configuration option, spelling replacement rule, and dictation history log resides exclusively on your local disk. We collect zero data. Match the table below to locate your secure app data paths on Windows 11 / 10:

Writing Custom Style Snippets

Style snippets let you apply a named formatting prompt to any dictation using a single spoken cue. The instructions are stored as standard JSON in your local configuration files.

{
  "email": "Format this transcript into a professional corporate email...",
  "bullets": "Summarize the key action items in clean, prioritized bullet points.",
  "refactor": "Read the spoken instructions and output only clean code..."
}

To invoke a snippet during a dictation session, simply click the active filter tag inside the capsule settings dashboard, or verbally prefix your speech with the snippet's name trigger.

Phonetic Vocabulary Correction

Tired of AI engines mishearing names, jargon, or command names? Productive Voice's custom dictionary intercepts raw phonetics before they are finalized. You teach it the sound-alike spelling, and Productive Voice automatically outputs the case-sensitive correct target.

"cuber nettys"  -->  "Kubernetes"
"postgres"      -->  "PostgreSQL"
"tanaka san"    -->  "Tanaka-san"

This replacement happens locally inside your thread loops in milliseconds, ensuring consistent accuracy even when working with highly specific technical files, namespaces, or native industry slang.

Windows Troubleshooting & Diagnostics

If Productive Voice fails to paste transcripts or trigger recordings, verify these three essential Windows platform access controls:

1. Microphone Access Blocked: Navigate to Settings -> Privacy & security -> Microphone. Ensure "Microphone access" is toggled ON and that desktop apps are permitted to access your recording hardware.
2. Focus Paste Outages: Verify that your target application runs with standard user focus. If target software runs with Administrator privileges (e.g. an admin cmd terminal), Productive Voice must also be launched as Administrator to simulate paste keyboard inputs.
3. API Endpoint Latencies: Check Port 443 outbound routing. High network jitter can cause socket dropouts. You can test your network health and key binds using our native Compatibility Scanner terminal widget.