Time & Temp Announcer with Asterisk

I have a WeatherFone Time and temp announcer system that no longer works and needs to be replaced what it does is when someone calls the phone number it gives them a recorded message with an advertisement them the time, temperature and short whether forecast. I was wondering if this would be possible to do with Asterisk. I would like the voice to be a recorded voice not a computer generated robotic voice. If you know how this could be done or have any ideas please let me know.


Try ‘googling’ for ‘nerdvittles weather asterisk’

The advertisement message can be played using Playback, the Weather part most of the time is grabbed from external web service and played using TTS.

Before you discard TTS, you should try IBM’s Watson or Amazon’s Polly. We’re so happy with IBM’s Watson we’ve dropped all of our English speaking voice talent.

As soon as we figure out how to convince Watson to ‘switch languages’ in the middle of a sentence so company names are pronounced correctly we’ll drop our Spanish voice talent as well.

So how would I set that up so it say the time and temp and them the weather?

You concatenate the strings and pass them to Watson using cURL. Watson renders the text to audio which you save as a file. You pass the file names(s) to Asterisk to play them.

It can take several seconds for Watson to respond based on the length of the string, so you’ll probably want to send each string (ad, weather, date & time) separately and cache the responses so you only have to render each string when it changes.

You also will want to cache the text since Watson charges by the character (IIRC). This may sound expensive, but you get a bunch (millions) as a ‘free’ baseline per month and only pay for overages. We only pay a couple of bucks every couple of months when we go over.

Sometimes you have to play with the text so it is pronounced correctly. Like ‘record’ is pronounced like a phonograph record and ‘re cord’ is pronounced like what you do with a microphone.

Here’s a sample:

watson.wav.txt (125.3 KB)

Do you know if this would be possible to do with the 3CX phone systems call flow designer?

No clue. Maybe a 3cx forum could answer that question.

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.