The talking boat project had its first public outing tonight at the second Bristol Raspberry Jam at the Bristol and Bath Science park, hosted again by Haber. Technical delays on the other talks got me put on first and I have to say the RPi worked well. Better than its presenter in fact, who messed up the buttons sometimes. My idea was to get the RPi to interview me, using the text reader functionality to step through some questions. Lucky there were two mics, one each.
Some lessons learnt: The GPS code needs to figure out when it has lost signal so it doesnt overwrite a good position with rubbish. The loss of position inside the building meant that demonstrating relative positioning would not have worked. It also lost its time reference after a reboot, having neither the internet nor GPS to sync to (actually I havent figured out how to sync unix to the GPS anyway - may just have to do that in the python code).
Arthur Amarra demoed his voice-controlled lamp. I'm not convinced the RPi is up to this task and I'm glad I stuck to a simple wireless keypad device like the presenter, although it too has its moments. At least I'm happy to let Arthur sort out speech recognition first!
Phil Atkin demoed his impressive synthesiser - facing complexity and timing problems far in excess of my talking boat. He pointed me to a USB sound I/O device which would improve the quality of sound I'm getting form the audio socket - but surely that should produce better sound than it does? Perhaps I should brave up to updating the firmware and perhaps I need another board, now RAM is up to 512Mb for the same price and those polyfuses have gone.
Very nice to see some friends there but numbers were down on the first meeting, and very few from schools. A pity generally and personally since I think the use of text-to-speech via espeak opens up the possibility of some simple yet rewarding projects on the Raspberry Pi.
Although the GPS didnt work in the building, it logged the track from home to the meeting and back, and told me speed and direction on the way back. The temporal resolution is only 20 seconds, hence corners get cut, but the spatial resolution is good considering the receiver is inside the car. Now I'd like to figure out is how to vary the line width with speed.