Python script for MacOS that speaks interjections from Historic Hansard. The interjections, voice, speaking rate, volume, and delay are all set randomly.

speak.py

import subprocess
from pymongo import MongoClient
import time
import random
import argparse

from credentials import MONGO_URL

HOUSES = {
    'hofreps': 'House of Representatives',
    'senate': 'Senate'
}


def get_interjections():
    dbclient = MongoClient(MONGO_URL)
    db = dbclient.get_default_database()
    match = {'length': {'$lte': args.max_length}}
    if args.query:
        match['$text'] = {'$search': args.query}
        interjections = list(db.interjections.aggregate(
            [
                {'$match': match},
                {'$sample': {'size': int(args.number)}},
            ]))
    else:
        interjections = list(db.interjections.aggregate(
            [
                {'$sample': {'size': int(args.number)}},
                {'$match': match}
            ]))
    return interjections


def get_voices():
    # This is MacOS specific stuff.
    voices = []
    # Get details of all the currently installed system voices
    system_voices = subprocess.check_output(['say', '-v', '?'])
    for voice in system_voices.split('\n'):
        parts = voice.split(None, 2)
        try:
            # Add all English speaking voices to the list of voices
            if parts[1][:2] == 'en':
                voices.append(parts[0])
        except IndexError:
            pass
    return voices


def speak():
    # Again very MacOS specific
    interjections = get_interjections()
    for interjection in interjections:
        text = interjection['text'].encode('utf-8')
        # Choose a random voice
        voice = random.choice(voices)
        print '\n\n{} (as voiced by {}), {}, {}:\n'.format(interjection['speaker']['name'].upper(), voice, HOUSES[interjection['house']], interjection['date'].strftime('%-d %B %Y'))
        print "    '{}'".format(text)
        # Set the volume somewhere between 50% & 100%
        subprocess.call(['osascript', '-e', 'set volume output volume {}'.format(random.randrange(50, 100, 10))])
        # Speak! Note randomly selected speaking rate between 150 and 250 words per minute
        subprocess.call(['say', '-v', voice, text.replace('"', '').replace("'", ""), '-r', str(random.randrange(150, 250, 25))])
        # Random delay between interjections from 0 to 5 seconds
        time.sleep(random.randrange(0, 5, 1))


# To run: python speak.py -n [number of tweets] -q [search word or phrase] -l [maximum length of interjection]
# For example: python speak.py -n 20 -q "\"White Australia\"" -l 200
parser = argparse.ArgumentParser()
parser.add_argument('-n', '--number', type=int, default=20, help='Number of interjections')
parser.add_argument('-q', '--query', help='Filter results by word or phrase')
parser.add_argument('-l', '--max_length', type=int, default=200, help='Maximum length of interjection')
args = parser.parse_args()
voices = get_voices()
speak()