search

krogenth / OpenShock.VoiceRecognizer

Voice Recognized zappies
MIT License
1 stars 0 forks source link
openshock

readme

OpenShock.VoiceRecognizer

Disclaimers

  1. The Browser Proxy settings currently only function on Windows, looking for specific registry keys to launch the specified browser.
  2. By default, the Browser Proxy will capture audio from your DEFAULT microphone. Please be aware of this.

Setup

Setup assumes you are wanting to others' voices to trigger shocks. If you are only looking for your own voice to trigger shocks you may simply capture your microphone capture device.

  1. Setup VoiceMeeter Banana or VoiceMeeter Potato.
  2. If you want to use local speech recognition, you can download a Vosk model from here. Extract the model preferably local to OpenShock.VoiceRecognizer. Try different models to see what works best for you and your hardware.
  3. If you want to use your browser (Chrome or Edge currently), then there is no additional software to acquire.
  4. Whatever program you are wanting to capture audio from, try to keep it on a separate output audio device (you will want to capture the capture device this is played to). If there are multiple programs, then they can all be on the same output device.
  5. In Banana or Potato, play your output device to a single virtual input device (do not have multiple output devices to this input to keep background noise to a minimum).
  6. Now you may run OpenShock.VoiceRecognizer.exe. Open the Options > Settings menu, and customize the settings as you see fit.
  7. Ensure the General > Input is set to whatever virtual (or phsyical) capture device you used in banana or potato.
  8. For OpenShock ensure you have an API Key you can provide to the program to be able to communicate to send commands. This can be set in the OpenShock sedttings.
  9. If you are using a Browser Proxy, ensure you set the Port to a non-zero value (recommended 5000, but use whatever is available for you).
  10. If you are using Vosk, set the Vosk > Model setting to the base directory of the extracted model from step #2.
  11. Once all settings have been modified, press Save and then close the window via the Close button or x.
  12. Now you can press the main Start button. If you are using a Browser Proxy, a new tab in the browser chosen will open. Follow the instructions in the browser tab to begin capturing sound data.

OSC

There are two available OSC endpoints:

  1. /recognizer/toggle: toggles the current active state of the recognizer.
  2. /reocgnizer/set: sets the current state of the recognizer explicitly. This takes a bool value.