Platform independent audio output. Stop hardcoding /home/jasper.

[astahlman]

Summary

This commit includes changes to support platform independence. This implementation detects the current platform and chooses an appropriate system command for audio output. It also does not rely on a directory structure rooted at /home/pi

Changes

1. Use $JASPER_HOME as base directory instead of assuming that root is /home/pi. This change is backwards-compatible, as the boot script will export $JASPER_HOME=/home/pi if this directory exists. On OS X, I set this environment variable in my ~/.zshrc.
2. Abstracted out the 'say' functionality into the new speaker module. The speaker module will select one of two implementations based on the detected platform: RPi, which still uses espeak, or OS X, which uses the built-in say command.
3. Updated boot/test.py to validate the list of vocabulary words passed to g2p without respect to ordering.

Testing

• Manual verification on both OS X and the RPi.
• Unit tests in boot/test.py and client/test.py pass on the RPi.

Notes

I wanted to minimize the scope of the changes included in this pull request, so I didn't mess with the directory structure. I was uneasy, however, appending client/ to the path in order to import the speaker module from inside boot/boot.py. What do others think should be the right approach here?

[vicramon]

The tests all pass on my Mac (10.9.3), but Japser is not recognizing any audio input once I boot it up. Did you do anything special to configure your microphone @astahlman?

[astahlman]

Hmmm, weird. No, the microphone didn't need any extra set up for me. (10.7.5)

[charliermarsh]

Nit: can this be named UnorderedList?

[charliermarsh]

This is good stuff. I still need to test it out, but I've left some comments above. Hopefully we're not too far away from a merge here. Thanks!

[astahlman]

Thanks. All good suggestions @crm416 - see the latest diff.

[charliermarsh]

Thanks for the quick turnaround. @shbhrsaha and I will try to test this before the end of the weekend.

[wadefletch]

Speaking of Platform-Independance, I have almost everything except audio input (thus requiring the use of local_mic) working on the Pythonista iOS app, but my TTS is a female voice named Jasper. Theoretically I could use an online TTS engine, but it would increase latency dramatically. If anyone knows about audio input on Pythonista or better iOS TTS it would be greatly appreciated.

[shbhrsaha]

This is great work, @astahlman .

Just one thing: on my Pi, your implementation works if I run boot.sh manually, but doesn't seem to work when Jasper boots via cron. main.py simply doesn't start.

I believe the problem lies in the bash conditional statements in boot.sh. To diagnose this, I removed the conditional statements and forced the JASPER_HOME variable to be set, after which main.py started up fine during boot.

Unfortunately I don't have a monitor so can't provide more cron output, and my limited Bash/cron knowledge can't offer a specific solution. Can you confirm that your implementation works on the Pi as it boots, in addition to running boot.sh manually?

[astahlman]

Good catch, @shbhrsaha. I disabled the automatic startup on my Pi so I didn't notice during testing. See the latest commit - I added a shebang at the beginning of the script to designate bash as the interpreter. Jasper now starts up successfully on reboot of the Pi.

[shbhrsaha]

Excellent. I'll get a chance to test this out in the next few days and will report back and merge this in if it's all good!

[jessecooper]

I did this to my fork as well. I use $HOME for the system calls and in the shell scripts.

[shbhrsaha]

Great, thanks! Merged into master and your contribution acknowledged in the README