def get_praat_pitch(audio_file):
"""Given a wav file, use Praat to return a dictionary containing pitch (in Hz)
at each millisecond."""
pl = PraatLoader(praatpath=path.abspath('Praat.app/Contents/MacOS/Praat'))
praat_output = pl.run_script('pitch.praat', audio_file)
pitch_data = pl.read_praat_out(praat_output)
return pitch_data
def get_praat_pitch(audio_file):
"""Given a wav file, use Praat to return a dictionary containing pitch (in Hz)
at each millisecond."""
pl = PraatLoader(praatpath=path.abspath('Praat.app/Contents/MacOS/Praat'))
praat_output = pl.run_script('pitch.praat', audio_file)
pitch_data = pl.read_praat_out(praat_output)
return pitch_data
def praat_analyze_pitch(audio_file):
"""Given a wav file, use Praat to return a dictionary containing pitch (in Hz)
at each millisecond."""
praatpath = path.abspath('Praat.app/Contents/MacOS/Praat') # locate Praat executable
pl = PraatLoader(praatpath=praatpath) # create instance of PraatLoader object
praat_output = pl.run_script('pitch.praat', audio_file) # run pitch script in Praat
pitch_data = pl.read_praat_out(praat_output) # turn Praat's output into Python dict
return pitch_data
def _檢查查甫查某(cls, 檔名):
pl = PraatLoader()
音懸結果 = pl.run_script('pitch.praat', 檔名)
查某 = 0
查甫 = 0
# print(檔名, 音懸結果)
for 資料 in pl.read_praat_out(音懸結果).values():
音懸 = 資料['Pitch']
if 音懸 != 0:
if 音懸 > cls.查某查甫音懸門檻:
查某 += 1
else:
查甫 += 1
# 一個0.001秒
if 查某 + 查甫 <= 50:
return None
if 查某 >= 查甫:
return '查某'
return '查甫'
from praatinterface import PraatLoader
pl = PraatLoader(praatpath = '/usr/bin/praat')
in_dir = '/home/pi/Documents/Project_Jammin/speech_emotion_classifier/wav/wav/carlos/casa_may_24_2016'
#in_dir = '/home/pi/Documents/Project_Jammin/speech_emotion_classifier/test'
out_dir = '/home/pi/Documents/Project_Jammin/speech_emotion_classifier/vectors/carlos/casa_may_24_2016/mfcc'
#out_dir = '/home/pi/Documents/Project_Jammin/speech_emotion_classifier/vectors/live'
text = pl.run_script('/home/pi/Documents/Project_Jammin/speech_emotion_classifier/wav2mfcc.praat', in_dir, out_dir )
#text = pl.run_script('/home/pi/Documents/Project_Jammin/speech_emotion_classifier/formants.praat','/home/pi/Documents/Project_Jammin/speech_recognition/wav/test.wav', 5500, 5)
print text
mats = pl.read_praat_out(text)
print mats
print "Done"
# define important directories here
trainpath = '/home/ashu/Desktop/emotion/hack-the-talk-exotel/training_dataset'
temppath = '/home/ashu/Desktop/emotion/hack-the-talk-exotel/temp'
praatpath = '/usr/bin/praat'
sound = AudioSegment.from_mp3("one.mp3")
print('imported...')
# slicing necessary as program gets stuck with long files
# we can find out optimal length and then work with it in a loop
# lets extend this script to extract features for all files and save them
# in some format
song = sound[:1000]
song.export("one.wav", format="wav")
print('exported...')
praatpath = '/usr/bin/praat'
pl = PraatLoader(praatpath = praatpath, debug = True)
text = pl.run_script('formants.praat', '/home/ashu/Desktop/emotion/hack-the-talk-exotel/features/one.wav', 5, 5500)
formants = pl.read_praat_out(text)
for x in formants:
print (x)
for y in formants[x]:
print (y,':',formants[x][y])
print(type(formants))
# formants is a dictionary so we can append it to create dictionary for whole time interval
# this way we have several dictionaries for several features
# we can also manipulate these dictionaries to form arrays (lists) for corresponding feature
def test_formants(praatpath, sound_file):
pl = PraatLoader(praatpath = praatpath, debug = True)
text = pl.run_script('formants.praat',sound_file, 5, 5500)
formants = pl.read_praat_out(text)
def test_formants(praatpath, sound_file):
pl = PraatLoader(praatpath=praatpath, debug=True)
text = pl.run_script('formants.praat', sound_file, 5, 5500)
formants = pl.read_praat_out(text)