I still don't know what to call this.
(http://i.imgur.com/u7smQQ6.png)
It'll be some kind of information terminal accessible over 70cm amateur radio, using both text-to-speech and speech-to-text. So far, I have code to get the current weather conditions written for it, well before I finished any of the backend components. It's designed for Orange Pi One (it's cheap \o/), but can work on normal RasPi or even Windows! Once I can tell that code is at least marginally functional, it'll be open-source on my github page. I also need to make sure that the Orange Pi One can do what I need it to with its limited USB, since I need both a sound card and wireless networking.
I want this project to be:
- flexible - work on many different platforms and radios,
- extensible - allow users to add functionality,
- simple - use as few hacks as possible,
- inexpensive - entire build under $55, and
- legal - when operating on the air, reading all that FCC legalese is necessary.
STATUS:- Audio Flow: 12%
- GPIO: 75%
- Radio Control: 0%
- Power Monitoring: 0%
- Voice Input: 0%
- Voice Output: 50%
- Weather: 100%
- Date and Time: 0%
Hm interesting. I am curious about how the audio quality will be. Also by text to speech, do you mean you'll code a converter?
Quote from: DJ Omnimaga on July 22, 2016, 02:08:59 AM
Hm interesting. I am curious about how the audio quality will be.
Bad, but copiable.
Quote from: DJ Omnimaga on July 22, 2016, 02:08:59 AM
Also by text to speech, do you mean you'll code a converter?
I'm probably going to be using Google for text-to-speech and speech-to-text (https://pypi.python.org/pypi/SpeechRecognition/). I don't have the time or knowledge to algorithmically process audio. I'll train my code to accept common mistranslations; since radio audio is filtered, speech recognition engines will have a hard time with translating voice.
Moderator edit: fixed formatting
At least, for date and time, you can always use the WWV stations (or CHU in Canada), which are pretty much made to be easily understood by computers. Plus, with CHU, you may even hear the voice of Simon Durivage every minute. You might even probably be able to make a NTP server with that.
I'm just going to query the date/time on the Pi and send that as speechified text; I don't want to get into rebroadcasting laws.
Thanks for explaining about Text to Speech. I didn't know Google made such thing. I knew AT&T had a text to speech converter.