Bug 278598 – speechSynthesis has aggressive ducking on iOS Safari

• Status: NEW
• Component: Media

[awe.media]

Created attachment 472284 [details]
.zip file with html/js demo and mp3 file

On iOS17 Safari the speechSynthesis always ducks other audio and this persists for about a second after it finishes speaking.

This does not happen on OSX Safari or Chrome on other platforms.

navigator.audioSession.type does not fix this either.

Attached is a simple demo that shows this issue. Here's how to use this demo.

- load the demo.html page
- tap on the "Tap to start" screen to enable any permission required
- tap on the "Tap to play" to hear the mp3 audio sample play (you can do this as often as you want and it resets currentTime to 0)
- tap on the "Tap to speak" to hear speechSynthesis say something

What is expected?
At least with navigator.audioSession.type='ambient' or navigator.audioSession.type='record-and-play' the 2 different audio streams (speech and mp3) should be played over the top of each other, both at their full volume. This is what happens on OSX Safari and Chrome on other platforms.

What does happen?
On iOS Safari the speechSynthesis ducks the mp3 while it's speaking, and keeps it's volume low for about 1 second after it finished speaking.

NOTE: Wasn't sure what component to put this under as speechSynthesis and audioSession didn't have obvious homes, so put it under "Media".

[awe.media]

Note that speechSynthesis has the same ducking impact on webAudio based playback too.

[Radar WebKit Bug Importer]

<rdar://problem/135056987>