def __init__(self,
keys,
dir_name='/home/brihi16142/work2/processed_emovdb_disgust'):
self.keys = keys
self.path = os.path.join(os.path.dirname(os.path.realpath(__file__)),
dir_name)
self.fnames, self.text_lengths, self.texts = read_metadata(
os.path.join(self.path, 'transcript_bea.csv'))
preprocess(dir_name, self)
print('Generated mels and mags')
def find_matches(samples, sr=RATE):
"""Finds fingerprint matches for the samples."""
print('Finding matching fingerprints...')
samples = preprocess(samples, sr)
fingerprints = generate_fingerprints('Microphone',
samples,
sr=sr,
plot=False)
print('Looking up matches in database...')
matches = lookup_fingerprints(fingerprints)
print('{} Matches'.format(len(matches)))
if len(matches) == 0:
return None
mapper = {}
for f in fingerprints:
mapper[f.hash] = f.time
diffed_matches = []
for f in matches:
diffed_matches.append((f.song_id, f.time - mapper[f.hash]))
return diffed_matches
def read_audio_file(filename):
"""Read MP3 file into a Song object."""
if not os.path.isfile(filename):
print('{} does not exists'.format(filename))
exit(1)
# Read metadata from song
meta = TinyTag.get(filename)
# Read mp3 and save as tempoary wavfile
ext = os.path.splitext(filename)[1].replace('.', '')
song = AudioSegment.from_file(filename, ext)
tmp_path = './tmp_{}'.format(os.path.basename(filename))
song.export(tmp_path, format='wav')
# Read and delete tempory wavefile
samplerate, samples = wav.read(tmp_path)
os.remove(tmp_path)
samples = preprocess(samples, samplerate)
s = Song(filename, meta, samples, samplerate)
return s
sys.exit(0)
else:
dataset_file_path = os.path.join(datasets_path, dataset_file_name)
if not os.path.isfile(dataset_file_path):
url = "http://data.keithito.com/data/speech/%s" % dataset_file_name
download_file(url, dataset_file_path)
else:
print("'%s' already exists" % dataset_file_name)
print("extracting '%s'..." % dataset_file_name)
os.system('cd %s; tar xvjf %s' % (datasets_path, dataset_file_name))
# pre process
print("pre processing...")
lj_speech = LJSpeech([])
preprocess(dataset_path, lj_speech)
elif args.dataset == 'mbspeech':
dataset_name = 'MBSpeech-1.0'
datasets_path = os.path.join(os.path.dirname(os.path.realpath(__file__)),
'datasets')
dataset_path = os.path.join(datasets_path, dataset_name)
if os.path.isdir(dataset_path) and False:
print("MBSpeech dataset folder already exists")
sys.exit(0)
else:
bible_books = ['01_Genesis', '02_Exodus', '03_Leviticus']
for bible_book_name in bible_books:
bible_book_file_name = '%s.zip' % bible_book_name
bible_book_file_path = os.path.join(datasets_path,
bible_book_file_name)
sys.exit(0)
else:
dataset_file_path = os.path.join(datasets_path, dataset_file_name)
if not os.path.isfile(dataset_file_path):
url = "http://data.keithito.com/data/speech/%s" % dataset_file_name
download_file(url, dataset_file_path)
else:
print("'%s' already exists" % dataset_file_name)
print("extracting '%s'..." % dataset_file_name)
os.system('cd %s; tar xvjf %s' % (datasets_path, dataset_file_name))
# pre process
print("pre processing...")
lj_speech = LJSpeech([])
preprocess(dataset_path, lj_speech)
elif args.dataset == 'mbspeech':
dataset_name = 'MBSpeech-1.0'
datasets_path = os.path.join(os.path.dirname(os.path.realpath(__file__)),
'datasets')
dataset_path = os.path.join(datasets_path, dataset_name)
if os.path.isdir(dataset_path) and False:
print("MBSpeech dataset folder already exists")
sys.exit(0)
else:
bible_books = ['01_Genesis', '02_Exodus', '03_Leviticus']
for bible_book_name in bible_books:
bible_book_file_name = '%s.zip' % bible_book_name
bible_book_file_path = os.path.join(datasets_path,
bible_book_file_name)
""" Gives an answer. """
box = driver.find_element_by_id("qpAnswerInput")
box.send_keys(ans)
box.send_keys(Keys.RETURN)
if __name__ == "__main__":
driver = login()
enter_game(driver, input("Room name? "), int(input("Room number? ")),
input("Room password? "))
while True:
try:
block_recording(driver)
print("starting recording...")
data = audio.record(LEN)
audio.sd.wait() # block on the recording
print("processing...")
vol1, clip = audio.preprocess(data)
ans = main.find_song(vol1, clip, VERBOSE)
if audio.np.max(clip) == 128: # 0 is at 128 because of the scaling
print("Clip is silent. Are you sure loopback is working?")
answer(driver, ans)
except KeyboardInterrupt:
driver.quit()
exit()
except Exception as e:
print(e)
ans = input("quit driver?\n")
if len(ans) > 0 and ans[0] == "y":
driver.quit()
args = parser.parse_args()
if args.dataset == 'ljspeech':
dataset_file_name = 'LJSpeech-1.1.tar.bz2'
datasets_path = os.path.join(os.path.dirname(os.path.realpath(__file__)),
'datasets')
dataset_path = os.path.join(datasets_path, 'LJSpeech-1.1')
if os.path.isdir(dataset_path) and False:
print("LJSpeech dataset folder already exists")
sys.exit(0)
else:
dataset_file_path = os.path.join(datasets_path, dataset_file_name)
if not os.path.isfile(dataset_file_path):
url = "http://data.keithito.com/data/speech/%s" % dataset_file_name
download_file(url, dataset_file_path)
else:
#print(os.system('pwd'))
#os.system('ls -al')
print("'%s' already exists" % dataset_file_name)
print("extracting '%s'..." % dataset_file_name)
#os.system('cd %s; tar xvjf %s' % (datasets_path, dataset_file_name))
os.system('cd datasets')
os.system('tar xvjf %s' % (dataset_file_name))
# pre process
print("pre processing...")
lj_speech = LJSpeech([])
preprocess(dataset_path, lj_speech)
sys.exit(0)
else:
dataset_file_path = os.path.join(datasets_path, dataset_file_name)
if not os.path.isfile(dataset_file_path):
url = "http://data.keithito.com/data/speech/%s" % dataset_file_name
download_file(url, dataset_file_path)
else:
print("'%s' already exists" % dataset_file_name)
print("extracting '%s'..." % dataset_file_name)
os.system('cd %s; tar xvjf %s' % (datasets_path, dataset_file_name))
# pre process
print("pre processing...")
lj_speech = LJSpeech([])
preprocess(dataset_path, lj_speech)
elif args.dataset == 'mbspeech':
dataset_name = 'MBSpeech-1.0'
datasets_path = os.path.join(os.path.dirname(os.path.realpath(__file__)),
'datasets')
dataset_path = os.path.join(datasets_path, dataset_name)
if os.path.isdir(dataset_path) and False:
print("MBSpeech dataset folder already exists")
sys.exit(0)
else:
bible_books = ['01_Genesis', '02_Exodus', '03_Leviticus']
for bible_book_name in bible_books:
bible_book_file_name = '%s.zip' % bible_book_name
bible_book_file_path = os.path.join(datasets_path,
bible_book_file_name)
from audio import preprocess
voice = 'LJ'
preprocess(f'datasets/{voice}')
print("done")
sys.exit(0)
else:
dataset_file_path = os.path.join(datasets_path, dataset_file_name)
if not os.path.isfile(dataset_file_path):
url = "http://data.keithito.com/data/speech/%s" % dataset_file_name
download_file(url, dataset_file_path)
else:
print("'%s' already exists" % dataset_file_name)
print("extracting '%s'..." % dataset_file_name)
os.system('cd %s; tar xvjf %s' % (datasets_path, dataset_file_name))
# pre process
print("pre processing...")
lj_speech = LJSpeech([])
preprocess(dataset_path, lj_speech)
elif args.dataset == 'en_uk':
dataset_file_name = 'en_UK.tgz'
datasets_path = os.path.join(os.path.dirname(os.path.realpath(__file__)), '')
dataset_path = os.path.join(datasets_path, 'en_UK')
if os.path.isdir(dataset_path) and False:
print("en_UK dataset folder already exists")
sys.exit(0)
else:
dataset_file_path = os.path.join(datasets_path, dataset_file_name)
if not os.path.isfile(dataset_file_path):
url = "http://data.m-ailabs.bayern/data/Training/stt_tts/%s" % dataset_file_name
download_file(url, dataset_file_path)
else:
print("'%s' already exists" % dataset_file_name)
import os
import sys
import csv
import time
import argparse
import fnmatch
import librosa
import pandas as pd
from hparams import HParams as hp
from zipfile import ZipFile
from audio import preprocess
from utils import download_file
from datasets.np_speech import NPSpeech
parser = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument("--dataset", required=True, choices=['NPSpeech'], help='dataset name')
args = parser.parse_args()
args.dataset == 'NPSpeech':
dataset_name = 'NPSpeech-1.0'
datasets_path = os.path.join(os.path.dirname(os.path.realpath(__file__)), 'datasets')
dataset_path = os.path.join(datasets_path, dataset_name)
# pre process
print("pre processing...")
np_speech = NPSpeech([])
preprocess(dataset_path, np_speech)