If you didn’t know, you can use a Python web server and the gTTS library to perform HTTP requests to get the audio, backed up by the fact you can see the inspect menu (e.g. Raining Tacos) so you can reverse-engineer the Sound class and put that as a SoundId, if possible. (The empty-extension blob file is just a bunch of binary)
Comment what you think you can do trying to get a TTS sound! (A module for this will be possible, just don’t make too many requests)