A fun project that uses the OpenAI TTS (text-to-speech) model to read selected text on your Mac with a keyboard shortcut. This is instead of using the default macOS accessibility text reader, since OpenAI's text-to-speech sounds much better.
As of June 2024, the OpenAI TTS model is priced at $15 per 1 million characters processed.
Select some text on a webpage or in the Apple Notes app, for example, and press the keyboard shortcut you defined. A window will appear letting you know how much the request will cost. (You can easily disable this feature if you want.)
If you press continue, after a few seconds, the OpenAI-generated audio will start playing out of your system's default audio output and reading your text aloud.
I'm using a combination of Python and Apple Scripts, plus an Automator Workflow to tie it all together and turn it into a macOS "Quick Action". Here's the Automator Workflow:
-
If you don't already have one, sign up for an OpenAI account and generate an API key. See their quickstart guide for more info.
-
Install the OpenAI Python library
pip install openai -
Download both Python scripts
openai-text-to-speech.pyandopenai-tts-cost-estimator.pyand theOpenAI Text to Speech.zipfile containing the Automator Workflow from this repo. -
Open
openai-text-to-speech.pyand replaceYOUR_OPENAI_API_KEY_GOES_HEREwith your OpenAI API key. -
Place both Python scripts in
/usr/local/bin. -
Unzip the
OpenAI Text to Speech.zipfile and double click theOpenAI Text to Speech.workflowfile to install it as a System Quick Action, or right click and open it with Automator to modify it to your liking first.
-
Finally, set up a keyboard shortcut on your Mac by visiting System Settings > Keyboard > Keyboard Shortcuts... > Services. Expand the
Textlist and find theOpenAI Text to Speechwe just installed. Define your desired shortcut.
You can also run the Quick Action/Service by right clicking on any selected text, and selecting the action from the Services menu option:
