Tool that allows for voice interaction with OpenAI chat without need to type long, detailed requests.
It's an terminal application that captures audio from microphone and queries chat with transcripted message.
Note: Currently text-davinci-003 model is used as a chat backend. Unfortunately API for chatGPT is not yet provided by OpenAI.
To install bro_listen package via PyPI, execute following command:
pip install bro-listenNote: Consider creating a separate virtual environment if you want to avoid polluting system with dependencies of bro_listen package.
poetry is required to install the package from the source code. You can get it here
Once poetry is enabled in the system, to install the project and its dependencies, execute following command:
git clone https://github.com/mikkac/bro_listen.git && cd bro_listen && poetry installAfter installation, configuration file has to be provided.
Default location used by the application is $HOME/.config/bro_listen/config.toml.
One can use the default configuration file. The only thing that needs to be provided is OpenAI API key that can be generated here.
Application supports several configuration parameters.
voice_api- Voice recognition API. Currently only"vosk"is supported, but it's planned to also enable usage of Google Speech-To-Text and Azure Speech to text.language- language used byvoice_api. Currently available languages are listed in Vosk's documentation.enable_audio_response- chat's responses are only written to console by default. However, it's possible to vocalize them by setting this parameter totrue.device_id- ID of recording device. It should be commented out or completely removed from configuration if default device shall be used.
If package has been installed via PyPI, run following command:
bro_listenIf it was installed from source, start the application with command:
poetry run bro_listenor
poetry shell
bro_listenThis project is licensed under the LICENSE file that can be found here