Convert audio recordings to text via macOS Enhanced Dictation and Soundflower #

Update

Environment

Setup

  1. Run Soundflower installer twice

  2. System Preferences > Sound > set both input and output device to "Soundflower (2ch)"

  3. System Preferences > Keyboard > Dictation > set "Dictation:" to "On", check "Use Enhanced Dictation", and set microphone to "Soundflower (2ch)"

  4. System Preferences > Accessibility > Dictation > uncheck "Mute audio output while dictating"

Use

  1. Begin playback of audio file

  2. Press Fn twice in TextEdit to start converting audio to text

Notes

  1. To uninstall Soundflower, delete Soundflower.kext from /Library/Extensions (and /System/Library/Extensions if present), sudo kextcache --prune-staging, and restart.

  2. BlackHole (FOSS), Sound Siphon ($49), or Loopback ($99) can be substituted for Soundflower.

  3. Enhanced Dictation has been removed from Catalina. Therefore, all Dictation data (along with a slew of other information) is sent to Apple for processing:

    Mojave Catalina
  4. This regrettable change, coupled with an error message sometimes encountered in Mojave ("Dictation languages aren't available to download at this time. Try again later."), makes it all the more advisable to back up a copy of the Enhanced Dictation files once they are downloaded. They can be found in /System/Library/Speech/Recognizers/SpeechRecognitionCoreLanguages/ with filenames like en_US.SpeechRecognition, ja_JP.SpeechRecognition, etc.

Sources

Related

/mac | Jun 16, 2020


Subscribe or visit the archives.