For this particular use case, I decided to give up on using the Web Speech API and instead combined the [Edge TTS Universal](https://www.npmjs.com/package/edge-tts-universal) library with Media Session API. On the plus site, the quality of many of the Edge-TTS voices is arguably much better than those of the Web Speech API. However, a downside is that the Edge-TTS is not instantaneous like the Web Speech API and Edge-TTS requires larger texts to be broken down into multiple smaller requests to the API.