AXA YAP : Hold-to-speak dictation app powered by SecureGPT

Hold a hotkey, speak, release and your speech is transcribed and pasted at the cursor.

Requirements

• Windows 11
• Python 3.13+
• uv installed

Configuration

Option 1 : Manually set environment variables

Set the following environment variables:

# SecureGPT credentials
ONEACCOUNT_LOGIN_URL=https://onelogin.axa.com
MHUB_API_URL=https://api.se.axa-go.applications.services.axa-tech.intraxa/ago-m365-securegpt-hub-v1-vrs
MHUB_CLIENT_ID=<your-client-id>
MHUB_CLIENT_SECRET=<your-client-secret>
MHUB_API_VERSION=2025-04-01-preview

Optionally, you can set these as well:

# Hotkey combo to hold while speaking (default: Ctrl+Alt+Space)
AXA_YAP_HOTKEY=<ctrl>+<alt>+<space>

# OpenAI transcription model
AXA_YAP_TRANSCRIPTION_MODEL=gpt-4o-transcribe-2025-03-20

# Microphone device index (run uvx axa_yap --list-devices to find yours, leave empty for default)
# Pass the index of the device (e.g. "0")
AXA_YAP_MIC_DEVICE=

# Language hint for transcription (e.g. "en", "fr"). Default: "en"
# Doesn't do much
AXA_YAP_LANGUAGE=en

Option 2: Use a .env file

Create a .env file in whatever location you want, and specify its path in the environment variable AXA_YAP_DOTENV_PATH.

Usage

Run directly with uvx (no manual install needed):

uvx axa_yap

Important: Use uvx axa_yap --list-devices to find your microphone index if the default doesn't work, and set AXA_YAP_MIC_DEVICE accordingly.

CLI Options

uvx axa_yap --list-devices # List available audio input devices
uvx axa_yap --install      # Register app to start automatically on Windows login
uvx axa_yap --uninstall    # Remove app from Windows startup

Instructions

1. Hold the hotkey (default Ctrl+Alt+Space) : recording starts and an overlay appears.
2. Speak while holding the keys.
3. Release: audio is sent to SecureGPT for transcription.
4. The transcribed text is automatically pasted at your current cursor position.