so, there are two ways to make this better.<li>Just generate a random waveform (possibly, matching some hash of the message that contains the audio).</li><li>Download the audio and re-encode the waveform locally.</li>
Either of these is better than PCM-in-JSON.