- Posted on
Fri May 26, 2017 9:46 am
-
wideglidejrp
offline
-
- Posts: 555
- Joined: Jan 15, 2012
- Location: Danbury, CT
-
Been giving some thought to the potential for a WU Announcements Plugin option. I just saw a survey showing the #1 reason people use Alexa, Siri, etc. Is to play music, but #2 was to announce weather information. This seems quite natural, with voice becoming more dominant in our digital lives. The Ivona option in Nick Lagaros's Sonos plugin opened my eyes to how pleasant announcements can be, much more so than normal text to speech. Unfortunately, Ivona will be shut down at the end of June, to be replaced by Polly. However, Nick's plugin does not support Polly. Here are a few thoughts on an alternative approach. I preface this assuming Polly will be a free AWS service for single person use. Not certain that is true.
The main difference I envision with an announcement plugin compared to the current plugin is in the text construction. Here is what I would love to see:
1. User selectable tokens such as ...
Opening tokens: "Good morning", "Good afternoon", "Good evening", "Attention: Weather Alert"
Closing tokens: "Have a good day in Philadelphia", "Have a good one", "Weather alert over"
2. User selectable units of measure...
F, C, None. This is important because of the difference between text announcements and voice announcements. For text it is natural to read "High will be 70 F", but for voice the F sounds out of place.
Miles per hour, Knots, None. To instead of dash. "The wind will be 15 to 20", or "The wind will be 15 to 20 miles per hour".
There may be other similar weather terms which would be different in voice than in text.
3. The plugin would create a text string from the tokens and forecast or alert information. The string would be submitted to AWS/Polly, which would then deliver an MP3 to the Sonos plugin folder or other user specified folder.
4. An Action would execute the Sonos plugin to play the MP3. For those without Sonos, I am sure there are alternatives for playing the MP3.
I have no idea how to do the AWS/Polly part of this idea, but I think it would be worthwhile. The Ivona/Polly technology is highly customizable with different voices and styles. It produces very natural sounding voice announcements.