The old Azure demo has been shut down, and the new link has been updated. You can only generate 500 characters at a time.
How it works
Record the screen with OBS, play back the recognized speech with Azure in Chrome, capture the system audio with Blackhole and pass it on to OBS, then export it from OBS as a video file and extract the audio. This method captures only the sound the system plays, avoiding interference from the microphone.
What you need
Steps
- Open Audio MIDI Setup in System - Utilities, click the plus sign in the bottom left corner -
Create Multi-Output Device, and check the boxes for both the device that plays system audio and Blackhole;
- Open OBS, and in Settings-Audio-Global Audio Devices-Mic/Auxiliary Audio select
Blackhole;
- Add any file as the video content to record (OBS can’t record just audio);
- Prepare the text to be recognized, then open Azure, listen to it and make sure it’s correct;
- Start
recording, then clickplayin Azure; - After the audio finishes playing, click
stop recordingin OBS; - The audio in the recording you just made is the audio you want.
- (Optional) Use ffmpeg to extract the audio from the mp4:
ffmpeg -i $HOME/Downloads/input.mp4 -b:a 320K -vn $HOME/Downloads/output.mp3
-
Previous
The Purpose of the value Variable in AE Expressions -
Next
Quickly Create a Metaball Effect in AE