jasper.rst

Getting started

Jasper is a voice computing platfrom for Raspberry pi. By integrating piwho with jasper you can:

• Write user specific modules
• Play user specific playlists
• Add speaker recognition to your Pi projects

Prerequisites

• Pi 1 / Pi 2 / Pi 3 (not tested)
• working copy of jasper's master or dev branch
• piwho library installed

Note: Take a backup of your jasper project.

configuring jasper:

The following changes will extract the recordings from jasper.

• If you're using master branch:

create a new directory inside jasper :

$ cd ~/jasper/
$ mkdir recordings

open /jasper/client/mic.py and add the following imports:

import time
import datetime
import shutil
import os

go to end of the file and add this function:

def gettime(self):
    current = time.time()
    tformat = datetime.datetime.fromtimestamp(current).strftime('%d:%H:%M:%S')
    return tformat

find function activeListenToAllOptions() and in that function comment out the following code:

with tempfile.SpooledTemporaryFile(mode='w+b') as f:
    wav_fp = wave.open(f, 'wb')
    wav_fp.setnchannels(1)
    wav_fp.setsampwidth(pyaudio.get_sample_size(pyaudio.paInt16))
    wav_fp.setframerate(RATE)
    wav_fp.writeframes(''.join(frames))
    wav_fp.close()
    f.seek(0)
    return self.active_stt_engine.transcribe(f)

and add the following code:

with tempfile.NamedTemporaryFile(mode='w+b') as f:
    wav_fp = wave.open(f, 'wb')
    wav_fp.setnchannels(1)
    wav_fp.setsampwidth(pyaudio.get_sample_size(pyaudio.paInt16))
    wav_fp.setframerate(RATE)
    wav_fp.writeframes(''.join(frames))
    wav_fp.close()
    f.seek(0)
    # gettime() function will replace temp file name with a timestamp
    path = os.path.join('~/jasper/recordings', (self.gettime() + '.wav'))
     # this will paste recorded file into your recordings folder
    shutil.copyfile(f.name, path)
    return self.active_stt_engine.transcribe(f)

save the file, run Jasper and execute any module. You should see a .wav file inside recordings folder. Now you're ready to use recognition library with jasper.

• If you're using dev branch:

create a new directory inside jasper-client:

$ cd ~/jasper-client/
$ mkdir recordings

open jasper-client/jasper/mic.py and add the following imports:

import time
import datetime
import shutil
import os

go to end of the file and add this function:

def gettime(self):
    current = time.time()
    tformat = datetime.datetime.fromtimestamp(current).strftime('%d:%H:%M:%S')
    return tformat

then in _write_frames_to_file(self, frames) function:

@contextlib.contextmanager
   def _write_frames_to_file(self, frames):
       with tempfile.NamedTemporaryFile(mode='w+b') as f:
           wav_fp = wave.open(f, 'wb')
           wav_fp.setnchannels(self._input_channels)
           wav_fp.setsampwidth(int(self._input_bits/8))
           wav_fp.setframerate(self._input_rate)
           wav_fp.writeframes(''.join(frames))
           wav_fp.close()
           f.seek(0)
           yield f

after f.seek(0) add the following code:

# gettime() function will replace temp file name with a timestamp
path = os.path.join('~/jasper/recordings', (self.gettime() + '.wav'))
# this will paste recorded file into your recordings folder
shutil.copyfile(f.name, path)

save the file, run Jasper and execute any module. You should see a .wav file inside recordings folder. Now you're ready to use recognition library with jasper.

Modules

See examples/jasper/ directory