name | Voice notes |
setting | OPENAI API KEY |
icon | mic_none |
instructions | This is a plugin that uses the Whisper AI integration to transcribe audio into text with high fidelity. Great for taking notes while on mobile, etc. ![]() This plugin requires your own OpenAI API key. This means that you will be charged by OpenAI based on your usage. Get your API key and use the plugin settings page to enter the value before you start using the plugin. linkInstructionsCall the plugin using Ctrl-O or Cmd-O or the search function on mobile. Type in "Voice notes". After you call the plugin once, a note will be created (the note will be titled "Voice notes" and will be tagged with system/voicenotes). Whenever you want to record a new voice note, simply visit that note and use the button to start a new recording. Notes: Please note: recordings are first sent to a server where they are converted to mp3, then sent to OpenAI. No audio files are kept on the conversion server, and everything is encrypted (https). Also, there are limits to the file size accepted by Whisper, so note that longer recordings might be rejected. Development notes: please read about Transient Activation to learn about various issues this plugin might encounter, especially on mobile, especially in contexts where the window/embed is not in focus. One such example is that writing stuff to the clibpoard is a gated API, which means that it cannot happen if the user wasn't recently active on the page linkChangelogChangelog: November 7th, 2024: Fix an issue where longer recordings would not be processed (due to user not interacting with the page for a longer time) No longer add transcription to the clipboard |
Development roadmap
Opt into saving mp3 to note (might need extra API support)
Add GPT integration that summarizes/extracts tasks