, , , , , , , , , , ,

Arkwood has a long-standing hatred of the BBC, and in particular BBC News. ‘It’s all a pack of lies,’ he said on their coverage of current affairs. But he still wants to know about the weather, and anything to do with Scotland. So I wrote some Python code on our tiny Raspberry Pi computer that listens to the BBC News channel, and whenever anything is mentioned on these subjects I alert him by way of playing a sound on his headphones. Sounds far-fetched, but read on…

Here’s the code

import pygame
from speech import Speech
from storage import Storage
from time import sleep

speech = Speech()
storage = Storage()

# constants
NEWS_FILE = "news.txt"
KEYWORDS_WEATHER = ["weather", "sun", "rain", "cloud"]
KEYWORDS_SCOTLAND = ["scotland", "scottish"]

# set up alerts
alert_weather = pygame.mixer.Sound("161545__dasdeer__rain.wav")
alert_scotland = pygame.mixer.Sound("90093__alienxxx__piper-and-drums.wav")

# loop until armageddon
while True:
    # get news chat
    news_chat = Speech().speech_to_text('/home/pi/PiAUISuite/VoiceCommand/speech-recog.sh').lower()
    # store chat
    storage.write_timestamp(NEWS_FILE, news_chat)
    # play alert if keyword found
    for word in news_chat.split():
        if (word in KEYWORDS_WEATHER):
            storage.write_timestamp(NEWS_FILE, "[play rain]")
        elif (word in KEYWORDS_SCOTLAND):
            storage.write_timestamp(NEWS_FILE, "[play bagpipes]")

    # pause before next sample

Okay. So first we set up some constants, including keywords that Arkwood wants us to listen out for. Then we set up some sound files to play when the keywords are uttered by a BBC anchorman.

The code then loops, using Google’s Speech To Text service to sample the BBC News channel with a microphone pressed up against the TV speaker. We store each sample in a file for reference – here’s an example of some of the news we have captured:


2014-04-24 18:45:16: committed to his world destruction you can have peace
2014-04-24 18:45:26: under what circumstances would you speak again
2014-04-24 18:45:37: do when i'm worse than other
2014-04-24 18:45:48: normal blood sugar message
2014-04-24 18:45:59: zebras the very people are committed to r&r
2014-04-24 18:46:11: your mom to clarity as long as im
2014-04-24 18:46:21: and i midland
2014-04-24 18:46:31: actually mean never
2014-04-24 18:46:43: pension of tools
2014-04-24 18:46:53: could i think
2014-04-24 18:47:04: pullman from the us 60 state john kerry
2014-04-24 18:47:15: if they continue to get anywhere
2014-04-24 18:47:26: plenty of people here
2014-04-24 18:47:36: free cod
2014-04-24 18:47:46: 
2014-04-24 18:47:56: the characterization of the
2014-04-24 18:48:08: big risk in the sense that as you know better than anybody
2014-04-24 18:48:19: mr me some old animosity
2014-04-24 18:48:29: ione both sides
2014-04-24 18:48:41: when they weren't ceremoniously boots
2014-04-24 18:48:52: most infamous and brotherhood and in egypt
2014-04-24 18:49:03: 6842 years ago to their own app
2014-04-24 18:49:13: i could sing materials into gaza
2014-04-24 18:49:25: bison his people have to deal with monistat to skepticism
2014-04-24 18:49:35: give your
2014-04-24 18:49:47: ukrainian government says its fully synthetic in action
2014-04-24 18:49:58: based on the border with ukraine
2014-04-24 18:50:09: exponent natalia until labor is internet
2014-04-24 18:50:19: 12
2014-04-24 18:50:30: ration earlier in the morning today
2014-04-24 18:50:41: the checkpoint they want some
2014-04-24 18:50:53: we had president putin
2014-04-24 18:51:04: should i buy the cranium hoopla gang

Now we can check each sample to see if it contains our keywords. If we find anything to do with the weather then we use Pygame to play our rain sound file (the weather is invariably rain in the UK, even when the forecast says ‘sun’ or ‘cloud’). If we find that Scotland has been mentioned then we play the bagpipes file instead. Arkwood’s headphones are attached to the audio jack of our Raspberry Pi.

2014-04-24 19:25:43: sacramento weather today
2014-04-24 19:25:43: [play rain]

‘My God!’ my debauched Belgian friend said, ‘It actually works. I can ignore all the BBC’s fibs until I hear the sound of my favourite subjects’.

I got the sound files from Freesound:

And to get the sound output through the headphones I made use of the following article:
Disable HDMI Audio & Force 1/8″ Jack Audio?

Google’s Speech To Text service is invoked from our Speech class:

from subprocess import Popen, PIPE
class Speech(object):
    # converts speech to text
    def speech_to_text(self, filepath):
            # utilise PiAUISuite to turn speech into text
            text = Popen(['sudo', filepath], stdout=PIPE).communicate()[0]
            # tidy up text
            text = text.replace('"', '').strip()
            # debug

            return text
            print ("Error translating speech")

And we store BBC News chat to file using our Storage class:

import datetime

class Storage(object):

    # write to timestamp file
    def write_timestamp(self, filename, value):
        timestamp = datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S")

            with open(filename, "a") as timestamp_file:
                timestamp_file.write(timestamp + ': ' + value.strip() + '\n')
            print ("Error writing timestamp")

I had one question for Arkwood though. ‘Hey. How come you’re interested in Scotland so much, what with you being Belgian and all?’

My buddy gave me an angry glare then replied, ‘I like lochs. What’s it to you, piss face?’