def __init__(self):
self.__checkOS()
if self.__osSystem == 'win32':
print("Loading library...")
Vokaturi.load("./lib/open/win/OpenVokaturi-3-3-win32.dll")
if self.__osSystem == 'linux':
Vokaturi.load('./lib/open/linux/OpenVokaturi-3-3-linux64.so')
def analyzeFile(filepath, lineNum):
print("in analyzeFile 1")
Vokaturi.load("./lib/Vokaturi_mac.so")
(sample_rate, samples) = scipy.io.wavfile.read(filepath)
print("in analyzeFile 2")
buffer_length = len(samples)
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1: # mono
c_buffer[:] = samples[:] / 32768.0
else: # stereo
c_buffer[:] = 0.5 * (samples[:, 0] + 0.0 + samples[:, 1]) / 32768.0
voice = Vokaturi.Voice(sample_rate, buffer_length)
print("in analyzeFile 3")
voice.fill(buffer_length, c_buffer)
quality = Vokaturi.Quality()
emotionProbabilities = Vokaturi.EmotionProbabilities()
voice.extract(quality, emotionProbabilities)
print("in analyzeFile 4")
if quality.valid:
emotions = [
emotionProbabilities.happiness, emotionProbabilities.sadness,
emotionProbabilities.anger, emotionProbabilities.fear,
emotionProbabilities.neutrality
]
if max(emotions) == emotionProbabilities.happiness:
maxEmotion = "Happy"
elif max(emotions) == emotionProbabilities.sadness:
maxEmotion = "Sad"
elif max(emotions) == emotionProbabilities.anger:
maxEmotion = "Angry"
elif max(emotions) == emotionProbabilities.neutrality:
maxEmotion = "Neut"
else:
maxEmotion = "Afraid"
stats = (
"Happy: %.3f\tSad: %.3f\tAngry %.3f\tFear %.3f\tNeut %.3f" %
(emotions[0], emotions[1], emotions[2], emotions[3], emotions[4]))
print("in analyzeFile 5")
emotionFile = open("emotions", 'a')
print("in analyzeFile 6")
writeEmotions(emotionFile, maxEmotion + " " + stats, lineNum)
print("in analyzeFile 7")
emotionFile.close()
print("in analyzeFile 8")
else:
print("Not enough sonorancy to determine emotions")
def voice_analysis():
print("Loading library...")
Vokaturi.load("../SIM/libs/OpenVokaturi-2-1/lib/Vokaturi_mac.so")
print("Analyzed by: %s" % Vokaturi.versionAndLicense())
print("Reading sound file...")
file_name = "demo.wav"
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
print(" sample rate %.3f Hz" % sample_rate)
print("Allocating Vokaturi sample array...")
buffer_length = len(samples)
print(" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1: # mono
c_buffer[:] = samples[:] / 32768.0
else: # stereo
c_buffer[:] = 0.5 * (samples[:, 0] + 0.0 + samples[:, 1]) / 32768.0
print("Creating VokaturiVoice...")
voice = Vokaturi.Voice(sample_rate, buffer_length)
print("Filling VokaturiVoice with samples...")
voice.fill(buffer_length, c_buffer)
print("Extracting emotions from VokaturiVoice...")
quality = Vokaturi.Quality()
emotionProbabilities = Vokaturi.EmotionProbabilities()
voice.extract(quality, emotionProbabilities)
if quality.valid:
with Connection('amqp://*****:*****@localhost:5672//') as conn:
producer = conn.Producer(serializer='json')
producer.publish(
{
'Neutral': format(emotionProbabilities.neutrality, '.3f'),
'Happy': format(emotionProbabilities.happiness, '.3f'),
'Sad': format(emotionProbabilities.sadness, '.3f'),
'Angry': format(emotionProbabilities.anger, '.3f'),
'Fear': format(emotionProbabilities.fear, '.3f')
},
exchange=media_exchange,
routing_key='voice',
declare=[voice_queue])
# print ("Neutral: %.3f" % emotionProbabilities.neutrality)
# print ("Happy: %.3f" % emotionProbabilities.happiness)
# print ("Sad: %.3f" % emotionProbabilities.sadness)
# print ("Angry: %.3f" % emotionProbabilities.anger)
# print ("Fear: %.3f" % emotionProbabilities.fear)
else:
print("Not enough sonorancy to determine emotions")
voice.destroy()
def vokatori_fun(file_name):
#print("Loading library...")
Vokaturi.load(
"C:/Users/gokhalea/HackathonCodes/OpenVokaturi-3-0a/OpenVokaturi-3-0a/lib/open/win/OpenVokaturi-3-0-win64.dll"
)
#print("Analyzed by: %s" % Vokaturi.versionAndLicense())
#print("Reading sound file...")
#file_name = sys.argv[1]
#(sample_rate, samples) = scipy.io.wavfile.read(file_name)
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
#print(" sample rate %.3f Hz" % sample_rate)
#print("Samples:" % samples)
#print("Allocating Vokaturi sample array...")
buffer_length = len(samples)
#print(" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1: # mono
c_buffer[:] = samples[:] / 32768.0
else: # stereo
c_buffer[:] = 0.5 * (samples[:, 0] + 0.0 + samples[:, 1]) / 32768.0
#print("Creating VokaturiVoice...")
voice = Vokaturi.Voice(sample_rate, buffer_length)
#print("Filling VokaturiVoice with samples...")
voice.fill(buffer_length, c_buffer)
print("Extracting emotions from VokaturiVoice...")
quality = Vokaturi.Quality()
emotionProbabilities = Vokaturi.EmotionProbabilities()
voice.extract(quality, emotionProbabilities)
if quality.valid:
print("Neutral: %.3f" % emotionProbabilities.neutrality)
print("Happy: %.3f" % emotionProbabilities.happiness)
print("Sad: %.3f" % emotionProbabilities.sadness)
print("Angry: %.3f" % emotionProbabilities.anger)
print("Fear: %.3f" % emotionProbabilities.fear)
print(
"______________________________________________________________________________________________"
)
return emotionProbabilities
else:
print("Not enough sonorancy to determine emotions")
voice.destroy()
return None
def analyze(file_name):
# print ("Loading library...")
Vokaturi.load("../lib/Vokaturi_linux64.so")
# print ("Analyzed by: %s" % Vokaturi.versionAndLicense())
# print ("Reading sound file...")
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
# print (" sample rate %.3f Hz" % sample_rate)
# print ("Allocating Vokaturi sample array...")
buffer_length = len(samples)
# print (" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1: # mono
c_buffer[:] = samples[:] / 32768.0
else: # stereo
c_buffer[:] = 0.5 * (samples[:, 0] + 0.0 + samples[:, 1]) / 32768.0
# print ("Creating VokaturiVoice...")
voice = Vokaturi.Voice(sample_rate, buffer_length)
# print ("Filling VokaturiVoice with samples...")
voice.fill(buffer_length, c_buffer)
# print ("Extracting emotions from VokaturiVoice...")
quality = Vokaturi.Quality()
emotionProbabilities = Vokaturi.EmotionProbabilities()
voice.extract(quality, emotionProbabilities)
enabled = True
if quality.valid:
# print ("Neutral: %.3f" % emotionProbabilities.neutrality)
# print ("Happy: %.3f" % emotionProbabilities.happiness)
# print ("Sad: %.3f" % emotionProbabilities.sadness)
# print ("Angry: %.3f" % emotionProbabilities.anger)
# print ("Fear: %.3f" % emotionProbabilities.fear)
value = emotionProbabilities.anger
else:
value = 0
enabled = False
voice.destroy()
return enabled, value
def callVokaturi(fileName):
# Loading Vokaturi Mac
Vokaturi.load("/Users/nchao/Desktop/Yale Hacks/lib/Vokaturi_mac.so")
# Reading sound files (.wav)
file_name = "/Users/nchao/Desktop/Yale Hacks/" + fileName
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
# Allocating Vokaturi sample array
buffer_length = len(samples)
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1:
c_buffer[:] = samples[:] / 32768.0 # mono
else:
c_buffer[:] = 0.5 * (samples[:, 0] + 0.0 +
samples[:, 1]) / 32768.0 # stereo
# Creating VokaturiVoice and filling it with voice sample
voice = Vokaturi.Voice(sample_rate, buffer_length)
voice.fill(buffer_length, c_buffer)
# Extracting emotions from Vokaturi
quality = Vokaturi.Quality()
emotionProbabilities = Vokaturi.EmotionProbabilities()
voice.extract(quality, emotionProbabilities)
emoDict = {
"Neutral": emotionProbabilities.neutrality,
"Happy": emotionProbabilities.happiness,
"Sad": emotionProbabilities.sadness,
"Angry": emotionProbabilities.anger,
"Fear": emotionProbabilities.fear
}
# Finding main emotion in voice file
sortedVals = sorted(emoDict.values())[::-1]
stdDev = numpy.std(sortedVals)
emotions = []
for percentage in sortedVals:
if percentage > abs(max(sortedVals) - 1.5 * stdDev):
emotions += [
key for key, val in emoDict.items() if val == percentage
]
voice.destroy()
return emoDict
def sentiment_analysis(path=sys.argv[1]):
print("Loading library...")
Vokaturi.load("./OpenVokaturi-3-0-linux64.so")
print("Analyzed by: %s" % Vokaturi.versionAndLicense())
print("Reading sound file...")
file_name = path
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
print(" sample rate %.3f Hz" % sample_rate)
print("Allocating Vokaturi sample array...")
buffer_length = len(samples)
print(" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1:
c_buffer[:] = samples[:] / 32768.0 # mono
else:
c_buffer[:] = 0.5 * (samples[:, 0] + 0.0 +
samples[:, 1]) / 32768.0 # stereo
print("Creating VokaturiVoice...")
voice = Vokaturi.Voice(sample_rate, buffer_length)
print("Filling VokaturiVoice with samples...")
voice.fill(buffer_length, c_buffer)
print("Extracting emotions from VokaturiVoice...")
quality = Vokaturi.Quality()
emotionProbabilities = Vokaturi.EmotionProbabilities()
voice.extract(quality, emotionProbabilities)
if quality.valid:
print("Neutral: %.3f" % emotionProbabilities.neutrality)
print("Happy: %.3f" % emotionProbabilities.happiness)
print("Sad: %.3f" % emotionProbabilities.sadness)
print("Angry: %.3f" % emotionProbabilities.anger)
print("Fear: %.3f" % emotionProbabilities.fear)
out_dict = {}
out_dict['neutral'] = emotionProbabilities.neutrality
out_dict['happy'] = emotionProbabilities.happiness
out_dict['sad'] = emotionProbabilities.sadness
out_dict['angry'] = emotionProbabilities.anger
out_dict['fear'] = emotionProbabilities.fear
voice.destroy()
return out_dict
def extract_emotions(file_path):
Vokaturi.load(get_vokaturi_lib())
(sample_rate, samples) = scipy.io.wavfile.read(file_path)
buffer_length = len(samples)
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1:
c_buffer[:] = samples[:] / 32768.0 # mono
else:
c_buffer[:] = 0.5 * (samples[:, 0] + 0.0 +
samples[:, 1]) / 32768.0 # stereo
voice = Vokaturi.Voice(sample_rate, buffer_length)
voice.fill(buffer_length, c_buffer)
quality = Vokaturi.Quality()
emotionProbabilities = Vokaturi.EmotionProbabilities()
voice.extract(quality, emotionProbabilities)
# if quality.valid:
print("Neutral: %.3f" % emotionProbabilities.neutrality)
print("Happy: %.3f" % emotionProbabilities.happiness)
print("Sad: %.3f" % emotionProbabilities.sadness)
print("Angry: %.3f" % emotionProbabilities.anger)
print("Fear: %.3f" % emotionProbabilities.fear)
emotions = {
'neutrality': "%.3f" % emotionProbabilities.neutrality,
'happiness': "%.3f" % emotionProbabilities.happiness,
'sadness': "%.3f" % emotionProbabilities.sadness,
'anger': "%.3f" % emotionProbabilities.anger,
'fear': "%.3f" % emotionProbabilities.fear
}
voice.destroy()
return emotions
def main():
print("Loading library...")
Vokaturi.load("lib/open/win/OpenVokaturi-3-3-win64.dll")
print("Analyzed by: %s" % Vokaturi.versionAndLicense())
r = RedisManager(host=RedisConfig['host'],
port=RedisConfig['port'],
db=RedisConfig['db'],
password=RedisConfig['password'],
decodedResponses=RedisConfig['decodedResponses'])
sub = r.getRedisPubSub()
sub.subscribe(RedisConfig['newAudioPubSubChannel'])
for item in sub.listen():
print(item) # Test
if item['type'] == 'message':
newMsg = item['data']
print("New Msg: " + str(newMsg)) # Test
if not isinstance(newMsg, str):
newMsg = newMsg.decode()
audioID = newMsg
audioContent = r.hgetFromRedis(
key=audioID, field=RedisConfig['audioHsetB64Field'])
audioParams = r.hgetFromRedis(
key=audioID, field=RedisConfig['audioHsetParamsField'])
if audioContent:
if isinstance(audioParams, bytes):
audioParams = audioParams.decode('utf-8')
if isinstance(audioContent, bytes):
audioContent = audioContent.decode('utf-8')
audioContent = base64.b64decode(audioContent)
audioContent = ast.literal_eval(audioContent.decode('utf-8'))
audioParams = ast.literal_eval(audioParams)
audioEmotions = extractEmotionsFromAudioFile(
audioContent, audioParams)
print(audioEmotions) # Test
if not audioEmotions:
audioEmotions = RedisConfig['voidMsg']
r.publishOnRedis(channel=RedisConfig['VocalChannel'],
msg=str(audioEmotions))
r.hsetOnRedis(key=audioID,
field=RedisConfig['audioHsetVocalResultField'],
value=str(audioEmotions))
def analyzeAudio(filename):
print("Loading library...")
Vokaturi.load("lib/OpenVokaturi-3-3/lib/open/win/OpenVokaturi-3-3-win64.dll")
print("Analyzed by: %s" % Vokaturi.versionAndLicense())
print("Reading sound file...")
(sample_rate, samples) = scipy.io.wavfile.read(filename)
print(" sample rate %.3f Hz" % sample_rate)
print("Allocating Vokaturi sample array...")
buffer_length = len(samples)
print(" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1:
c_buffer[:] = samples[:] / 32768.0 # mono
else:
c_buffer[:] = 0.5*(samples[:,0]+0.0+samples[:,1]) / 32768.0 # stereo
print("Creating VokaturiVoice...")
voice = Vokaturi.Voice(sample_rate, buffer_length)
print("Filling VokaturiVoice with samples...")
voice.fill(buffer_length, c_buffer)
print("Extracting emotions from VokaturiVoice...")
quality = Vokaturi.Quality()
emotionProbabilities = Vokaturi.EmotionProbabilities()
voice.extract(quality, emotionProbabilities)
if quality.valid:
print("Neutral: %.3f" % emotionProbabilities.neutrality)
print("Happy: %.3f" % emotionProbabilities.happiness)
print("Sad: %.3f" % emotionProbabilities.sadness)
print("Angry: %.3f" % emotionProbabilities.anger)
print("Fear: %.3f" % emotionProbabilities.fear)
voice.destroy()
return emotionProbabilities
def emotion_recognition(speech):
#import library Vokaturi
sys.path.append("../api")
Vokaturi.load("../lib/Vokaturi_mac.so")
(sample_rate, samples) = scipy.io.wavfile.read(speech)
buffer_length = len(samples)
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1:
c_buffer[:] = samples[:] / 32768.0 # mono
else:
c_buffer[:] = 0.5 * (samples[:, 0] + 0.0 +
samples[:, 1]) / 32768.0 # stereo
voice = Vokaturi.Voice(sample_rate, buffer_length)
voice.fill(buffer_length, c_buffer)
#extracting emotions from speech
quality = Vokaturi.Quality()
emotionProbabilities = Vokaturi.EmotionProbabilities()
voice.extract(quality, emotionProbabilities)
if quality.valid:
'''print("Neutral: %.3f" % emotionProbabilities.neutrality)
print("Happy: %.3f" % emotionProbabilities.happiness)
print("Sad: %.3f" % emotionProbabilities.sadness)
print("Angry: %.3f" % emotionProbabilities.anger)
print("Fear: %.3f" % emotionProbabilities.fear)'''
emotion = emotionProbabilities.max
percent = emotionProbabilities.max.percent
voice.destroy()
return emotion, percent
# Paul Boersma 2017-01-25
#
# A sample script that uses the Vokaturi library to extract the emotions from
# a wav file on disk. The file has to contain a mono recording.
#
# Call syntax:
# python3 measure_wav_linux32.py path_to_sound_file.wav
import sys
import scipy.io.wavfile
sys.path.append("../api")
import Vokaturi
print("Loading library...")
Vokaturi.load("../lib/Vokaturi_linux32.so")
print("Analyzed by: %s" % Vokaturi.versionAndLicense())
print("Reading sound file...")
file_name = sys.argv[1]
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
print(" sample rate %.3f Hz" % sample_rate)
print("Allocating Vokaturi sample array...")
buffer_length = len(samples)
print(" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1:
c_buffer[:] = samples[:] / 32768.0
else:
c_buffer[:] = 0.5 * (samples[:, 0] + 0.0 + samples[:, 1]) / 32768.0
from pathlib import Path
import sys
from google.cloud import speech
from google.cloud.speech import enums, types
import scipy.io.wavfile
from .config import project_root
from .state import get_state
vokaturi_directory = project_root() / 'deps/Vokaturi'
sys.path.append(str(vokaturi_directory / 'api'))
import Vokaturi
Vokaturi.load(str(vokaturi_directory / 'OpenVokaturi-3-0-linux64.so'))
def read_speech(path):
import subprocess
output = str(Path(path).resolve().parent / 'file.wav')
subprocess.check_call(['sox', '|opusdec --force-wav {} -'.format(path), output])
return scipy.io.wavfile.read(output)
def get_sentiment(sample_rate, samples):
print('Sample rate %.3f Hz' % sample_rate)
print('Allocating Vokaturi sample array...')
buffer_length = len(samples)
print('%d samples, %d channels' % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1:
Affiliation: Robotics Laboratory, Japan Advanced Institute of Science and Technology, Japan
Project: CARESSES (http://caressesrobot.org/en/)
'''
import sys
import audioop
import math
sys.path.append("./api")
import Vokaturi
import numpy as np
from collections import deque
from pyaudio import PyAudio, paInt16
Vokaturi.load("./lib/open/linux/OpenVokaturi-3-0-linux64.so")
class VocalEmotionEstimatorEngine():
CHUNK = 2048
RATE = 16000
THRESHOLD = 1500
SILENCE_LIMIT = 1
PREV_AUDIO = 0.5
def __init__(self):
self.data = []
self.audio2send = []
self.rel = VocalEmotionEstimatorEngine.RATE / VocalEmotionEstimatorEngine.CHUNK
self.slid_win = deque(
maxlen=VocalEmotionEstimatorEngine.SILENCE_LIMIT * self.rel)
def __init__(self):
self.__checkOS()
if self.__osSystem == 'win32':
print("Loading library...")
Vokaturi.load("../lib/open/win/OpenVokaturi-3-3-win64.dll")
fileCount = 0
osLoaded = False
print("1 for linux32 ")
print("2 for linux64 ")
print("3 for mac32 ")
print("4 for mac64 ")
print("5 for win32 ")
print("6 for win64 ")
selection = input("Please select os version: ")
#while loop for selecting the os version
while (osLoaded == False):
if selection == '1':
print("Loading...")
Vokaturi.load(
"../OpenVokaturi-3-0/lib/open/linux/OpenVokaturi-3-0-linux32.so")
osLoaded = True
print("OS loaded")
elif selection == '2':
print("Loading...")
Vokaturi.load(
"../OpenVokaturi-3-0/lib/open/linux/OpenVokaturi-3-0-linux64.so")
osLoaded = True
print("OS loaded")
elif selection == '3':
print("Loading...")
Vokaturi.load(
"../OpenVokaturi-3-0/lib/open/macos/OpenVokaturi-3-0-mac32.dylib")
osLoaded = True
#
# For the sound file hello.wav that comes with OpenVokaturi, the result should be:
# Neutral: 0.760
# Happy: 0.000
# Sad: 0.238
# Angry: 0.001
# Fear: 0.000
import sys
import scipy.io.wavfile
sys.path.append("../api")
import Vokaturi
print("Loading library...")
Vokaturi.load("../lib/open/macos/OpenVokaturi-3-0-mac64.dylib")
print("Analyzed by: %s" % Vokaturi.versionAndLicense())
print("Reading sound file...")
file_name = sys.argv[1]
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
print(" sample rate %.3f Hz" % sample_rate)
print("Allocating Vokaturi sample array...")
buffer_length = len(samples)
print(" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1: # mono
c_buffer[:] = samples[:] / 32768.0
else: # stereo
c_buffer[:] = 0.5 * (samples[:, 0] + 0.0 + samples[:, 1]) / 32768.0
#
# For the sound file hello.wav that comes with OpenVokaturi, the result should be:
# Neutral: 0.760
# Happy: 0.000
# Sad: 0.238
# Angry: 0.001
# Fear: 0.000
import sys
import scipy.io.wavfile
sys.path.append("../api")
import Vokaturi
print ("Loading library...")
Vokaturi.load("../lib/open/linux/OpenVokaturi-3-3-linux32.so")
print ("Analyzed by: %s" % Vokaturi.versionAndLicense())
print ("Reading sound file...")
file_name = sys.argv[1]
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
print (" sample rate %.3f Hz" % sample_rate)
print ("Allocating Vokaturi sample array...")
buffer_length = len(samples)
print (" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1: # mono
c_buffer[:] = samples[:] / 32768.0
else: # stereo
c_buffer[:] = 0.5*(samples[:,0]+0.0+samples[:,1]) / 32768.0
#
# For the sound file example.wav that comes with OpenVokaturi, the result should be:
# Neutral: 0.426
# Happy: 0.001
# Sad: 0.506
# Angry: 0.005
# Fear: 0.063
import sys
import scipy.io.wavfile
sys.path.append("../api")
import Vokaturi
print("Loading library...")
Vokaturi.load("../lib/Vokaturi_linux_arm64.so")
print("Analyzed by: %s" % Vokaturi.versionAndLicense())
print("Reading sound file...")
file_name = sys.argv[1]
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
print(" sample rate %.3f Hz" % sample_rate)
print("Allocating Vokaturi sample array...")
buffer_length = len(samples)
print(" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1: # mono
c_buffer[:] = samples[:] / 32768.0
else: # stereo
c_buffer[:] = 0.5 * (samples[:, 0] + 0.0 + samples[:, 1]) / 32768.0
# Paul Boersma 2017-01-25
#
# A sample script that uses the Vokaturi library to extract the emotions from
# a wav file on disk. The file has to contain a mono recording.
#
# Call syntax:
# python3 measure_wav_linux64.py path_to_sound_file.wav
import sys
import scipy.io.wavfile
sys.path.append("../api")
import Vokaturi
print ("Loading library...")
Vokaturi.load("../lib/Vokaturi_win64.dll")
print ("Analyzed by: %s" % Vokaturi.versionAndLicense())
print ("Reading sound file...")
file_name = sys.argv[1]
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
print (" sample rate %.3f Hz" % sample_rate)
print ("Allocating Vokaturi sample array...")
buffer_length = len(samples)
print (" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1: # mono
c_buffer[:] = samples[:] / 32768.0
else: # stereo
c_buffer[:] = 0.5*(samples[:,0]+0.0+samples[:,1]) / 32768.0
#
# For the sound file example.wav that comes with OpenVokaturi, the result should be:
# Neutral: 0.426
# Happy: 0.001
# Sad: 0.506
# Angry: 0.005
# Fear: 0.063
import sys
import scipy.io.wavfile
sys.path.append("../api")
import Vokaturi
print("Loading library...")
Vokaturi.load("../lib/Vokaturi_mac.so")
print("Analyzed by: %s" % Vokaturi.versionAndLicense())
print("Reading sound file...")
file_name = sys.argv[1]
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
print(" sample rate %.3f Hz" % sample_rate)
print("Allocating Vokaturi sample array...")
buffer_length = len(samples)
print(" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1: # mono
c_buffer[:] = samples[:] / 32768.0
else: # stereo
c_buffer[:] = 0.5 * (samples[:, 0] + 0.0 + samples[:, 1]) / 32768.0
# Neutral: 0.760
# Happy: 0.000
# Sad: 0.238
# Angry: 0.001
# Fear: 0.000
import sys
import scipy.io.wavfile
import Vokaturi
import os
vokaturi_path = os.path.dirname(os.path.realpath(__file__))
print(vokaturi_path)
print ("Loading library...")
Vokaturi.load(vokaturi_path + "/OpenVokaturi-3-0-mac64.dylib")
print ("Analyzed by: %s" % Vokaturi.versionAndLicense())
print ("Analyzed by: %s" % Vokaturi.versionAndLicense())
print ("Reading sound file...")
file_name = sys.argv[1]
(sample_rate, samples) = scipy.io.wavfile.read(vokaturi_path + "/"+ file_name)
print (" sample rate %.3f Hz" % sample_rate)
print ("Allocating Vokaturi sample array...")
buffer_length = len(samples)
print (" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1: # mono
c_buffer[:] = samples[:] / 32768.0
else: # stereo
# Angry: 0.001
# Fear: 0.000
import sys
import scipy.io.wavfile
sys.path.append("./")
import Vokaturi
os = sys.argv[2]
print("Loading library...")
# NOTE: This is working for OSX, Linux, Win64. If this doesn't work on your machine, you might have to install some other
# Vokaturi library like the win32.dll file and change the relative path accordingly
if os == 'windows':
Vokaturi.load("./OpenVokaturi-3-0-win64.dll")
elif os == 'linux':
Vokaturi.load("./OpenVokaturi-3-0-linux64.so")
elif os == 'mac' or os == 'osx' or os == 'macos':
Vokaturi.load("./OpenVokaturi-3-0-mac64.dylib")
print("Analyzed by: %s" % Vokaturi.versionAndLicense())
print("Reading sound file...")
file_name = sys.argv[1]
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
print(" sample rate %.3f Hz" % sample_rate)
print("Allocating Vokaturi sample array...")
buffer_length = len(samples)
print(" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
# Paul Boersma 2017-01-25
#
# A sample script that uses the Vokaturi library to extract the emotions from
# a wav file on disk. The file has to contain a mono recording.
#
# Call syntax:
# python3 measure_wav_mac.py path_to_sound_file.wav
import sys
import scipy.io.wavfile
sys.path.append("../api")
import Vokaturi
print ("Loading library...")
Vokaturi.load("../lib/Vokaturi_mac.so")
print ("Analyzed by: %s" % Vokaturi.versionAndLicense())
print ("Reading sound file...")
file_name = sys.argv[1]
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
print (" sample rate %.3f Hz" % sample_rate)
print ("Allocating Vokaturi sample array...")
buffer_length = len(samples)
print (" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1: # mono
c_buffer[:] = samples[:] / 32768.0
else: # stereo
c_buffer[:] = 0.5*(samples[:,0]+0.0+samples[:,1]) / 32768.0
import uuid
import gc
import sys
import scipy.io.wavfile
sys.path.append("api")
import Vokaturi
from shutil import copyfile
from confidence_prediction import test_example
import pickle
with open('logistic_regression_.pkl', 'rb') as f:
clf = pickle.load(f)
#from opus.decoder import Decoder as OpusDecoder
#Vokaturi.load("api/OpenVokaturi-3-3-win64.dll")
Vokaturi.load("api/OpenVokaturi-3-3-linux64.so")
print("Analyzed by: %s" % Vokaturi.versionAndLicense())
def analyze():
file_name = "tempFile.wav"
test_exm = test_example(file_name, clf)
return test_exm[0, 0], test_exm[0, 1]
clients = {}
class OpusDecoderWS(tornado.websocket.WebSocketHandler):
def open(self):
print('new connection')
#
# For the sound file hello.wav that comes with OpenVokaturi, the result should be:
# Neutral: 0.760
# Happy: 0.000
# Sad: 0.238
# Angry: 0.001
# Fear: 0.000
import sys
import scipy.io.wavfile
sys.path.append("../api")
import Vokaturi
print("Loading library...")
Vokaturi.load("../bin/Debug/OpenVokaturi-3-0-win64.dll")
#Vokaturi.load("../lib/open/win/OpenVokaturi-3-0-win32.dll")
print("Analyzed by: %s" % Vokaturi.versionAndLicense())
print("Reading sound file...")
file_name = sys.argv[1]
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
print(" sample rate %.3f Hz" % sample_rate)
print("Allocating Vokaturi sample array...")
buffer_length = len(samples)
print(" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1: # mono
c_buffer[:] = samples[:] / 32768.0
else: # stereo
#
# Call syntax:
# python3 analyze_one_file.py path_to_sound_file.wav
import sys
import os
import scipy.io.wavfile
sys.path.append(os.path.join(sys.path[0],'api'))
import Vokaturi
file_name = sys.argv[1]
libPath = os.path.join(sys.path[0],'lib','Vokaturi_mac.so')
Vokaturi.load(libPath)
os.system('clear')
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
samples = samples[:] / 32768.0
buffer_length = len(samples)
c_buffer = Vokaturi.SampleArrayC(buffer_length)
# something weird with original sample code here.
# try just the first element in array.
for index in range(len(samples)):
c_buffer[index] = samples[index][0]
voice = Vokaturi.Voice(sample_rate, buffer_length)
emotionProbabilities = Vokaturi.EmotionProbabilities()
#
# For the sound file hello.wav that comes with OpenVokaturi, the result should be:
# Neutral: 0.760
# Happy: 0.000
# Sad: 0.238
# Angry: 0.001
# Fear: 0.000
import sys
import pyaudio
import numpy as np
sys.path.append("/OpenVokaturi/api")
import Vokaturi
Vokaturi.load("/OpenVokaturi/lib/open/win/OpenVokaturi-3-0-win64.dll")
CHUNKSIZE = 100000 # fixed chunk size
sample_rate = 44100
# initialize portaudio
p = pyaudio.PyAudio()
device_info = p.get_default_input_device_info()
#device_info = p.get_device_info_by_index(1)
channelcount = device_info["maxInputChannels"] if (
device_info["maxOutputChannels"] < device_info["maxInputChannels"]) else device_info["maxOutputChannels"]
stream = p.open(format=pyaudio.paInt16,
channels=channelcount,
rate=int(device_info["defaultSampleRate"]),
input=True,
#
# For the sound file hello.wav that comes with OpenVokaturi, the result should be:
# Neutral: 0.760
# Happy: 0.000
# Sad: 0.238
# Angry: 0.001
# Fear: 0.000
import sys
import scipy.io.wavfile
sys.path.append("../api")
import Vokaturi
print ("Loading library...")
Vokaturi.load("../lib/open/win/OpenVokaturi-3-3-win32.dll")
print ("Analyzed by: %s" % Vokaturi.versionAndLicense())
print ("Reading sound file...")
file_name = sys.argv[1]
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
print (" sample rate %.3f Hz" % sample_rate)
print ("Allocating Vokaturi sample array...")
buffer_length = len(samples)
print (" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1: # mono
c_buffer[:] = samples[:] / 32768.0
else: # stereo
c_buffer[:] = 0.5*(samples[:,0]+0.0+samples[:,1]) / 32768.0
#
# For the sound file example.wav that comes with OpenVokaturi, the result should be:
# Neutral: 0.760
# Happy: 0.000
# Sad: 0.238
# Angry: 0.001
# Fear: 0.000
import sys
import scipy.io.wavfile
sys.path.append("../api")
import Vokaturi
print("Loading library...")
Vokaturi.load("../lib/Vokaturi_win64.dll")
print("Analyzed by: %s" % Vokaturi.versionAndLicense())
print("Reading sound file...")
file_name = sys.argv[1]
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
print(" sample rate %.3f Hz" % sample_rate)
print("Allocating Vokaturi sample array...")
buffer_length = len(samples)
print(" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1: # mono
c_buffer[:] = samples[:] / 32768.0
else: # stereo
c_buffer[:] = 0.5 * (samples[:, 0] + 0.0 + samples[:, 1]) / 32768.0
def listen_me():
global text, duration
parser = argparse.ArgumentParser()
parser.add_argument('--filename', '-f', default='recording.wav')
args = parser.parse_args()
# 라이브러리 준비
Vokaturi.load("/home/pi/lib/piZero.so")
# 클라우드 스피치, 텍스트 자연어처리, tts 클라이언트 각각 초기화
client = CloudSpeechClient()
nlp_client = language.LanguageServiceClient()
tts_client = texttospeech.TextToSpeechClient()
pos_wavs = []
neut_wavs = []
neg_wavs = []
intro_wavs = []
pos_wavs.append(text_to_audio(tts_client, '진짜?', '0.wav'))
pos_wavs.append(text_to_audio(tts_client, '대박', '1.wav'))
pos_wavs.append(text_to_audio(tts_client, '우와', '2.wav'))
pos_wavs.append(text_to_audio(tts_client, '하하', '3.wav'))
neut_wavs.append(text_to_audio(tts_client, '응', '10.wav'))
neut_wavs.append(text_to_audio(tts_client, '그렇구나', '11.wav'))
neut_wavs.append(text_to_audio(tts_client, '그래서?', '12.wav'))
neut_wavs.append(text_to_audio(tts_client, '응응', '13.wav'))
neg_wavs.append(text_to_audio(tts_client, '저런', '4.wav'))
neg_wavs.append(text_to_audio(tts_client, '힘내', '5.wav'))
neg_wavs.append(text_to_audio(tts_client, '에휴', '6.wav'))
intro_wavs.append(text_to_audio(tts_client, '들어줄게. 얘기해봐', 'intro0.wav'))
intro_wavs.append(text_to_audio(tts_client, '무슨 일 이야?', 'intro1.wav'))
play_wav(random.choice(intro_wavs))
logging.basicConfig(level=logging.INFO)
with Board() as board:
while True:
print('말해보자.')
text = None
duration = 0.
emotion = None
def wait():
global text, duration
start = time.monotonic()
while text is None:
# 텍스트로 인식
text = client.recognize(language_code='ko-KR')
duration = time.monotonic() - start
# 녹음하면서
record_file(AudioFormat.CD,
filename=args.filename,
wait=wait,
filetype='wav')
print(text)
print('Recorded: %.02f seconds' % duration)
if text in ['들어줘서 고마워', '내 얘기 들어줘서 고마워', '어시스턴트', '잘가', '잘 가']:
return
# 텍스트 감정 분석
document = types.Document(content=text,
type=enums.Document.Type.PLAIN_TEXT)
sentiment = nlp_client.analyze_sentiment(
document=document).document_sentiment
print('텍스트 감정 분석*********************************')
print('Text: {}'.format(text))
print('Sentiment: {}, {}'.format(sentiment.score,
sentiment.magnitude))
##################### 실험후 바꿔도 됨 ####################
pos_standard = 0.6
neg_standard = 0.1
# magnitude_standard = 0.1
# text sentiment analysis is enough
if (sentiment.score < neg_standard
or sentiment.score > pos_standard):
if sentiment.score < neg_standard:
emotion = False
print("@@@negative")
else:
emotion = True
print("@@@positive")
else:
# 녹음 파일 감정 분석
print('오디오 감정 분석*********************************')
(sample_rate, samples) = scipy.io.wavfile.read(args.filename)
# print (" sample rate %.3f Hz" % sample_rate)
# print ("Allocating Vokaturi sample array...")
buffer_length = len(samples)
print(" %d samples, %d channels" %
(buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1: # mono
c_buffer[:] = samples[:] / 32768.0
else: # stereo
c_buffer[:] = 0.5 * (samples[:, 0] + 0.0 +
samples[:, 1]) / 32768.0
# print ("Creating VokaturiVoice...")
voice = Vokaturi.Voice(sample_rate, buffer_length)
# print ("Filling VokaturiVoice with samples...")
voice.fill(buffer_length, c_buffer)
# print ("Extracting emotions from VokaturiVoice...")
quality = Vokaturi.Quality()
emotionProbabilities = Vokaturi.EmotionProbabilities()
voice.extract(quality, emotionProbabilities)
if quality.valid:
# print ("Neutral: %.3f" % emotionProbabilities.neutrality)
# print ("Happy: %.3f" % emotionProbabilities.happiness)
# print ("Sad: %.3f" % emotionProbabilities.sadness)
# print ("Angry: %.3f" % emotionProbabilities.anger)
# print ("Fear: %.3f" % emotionProbabilities.fear)
# fear 는 무시하도록 하자.
wave_score = emotionProbabilities.happiness - (
emotionProbabilities.sadness +
emotionProbabilities.anger)
if wave_score > 0 and sentiment.score > 0.4:
print('@@@긍정')
emotion = True
elif wave_score < 0 and sentiment.score < 0.4:
print('@@@부정')
emotion = False
# text 스코어와 wave 스코어가 불일치 할때는 중립반응 (emotion = None)
# 여기서 부터 반응.
with Leds() as leds:
if emotion is True:
play_wav(random.choice(pos_wavs))
leds.pattern = Pattern.blink(100)
color = (255, 255, 0)
leds.update(Leds.rgb_pattern(color))
time.sleep(1)
# play_wav('laugh.wav')
elif emotion is False:
play_wav(random.choice(neg_wavs))
leds.pattern = Pattern.breathe(1000)
color = (102, 140, 255)
leds.update(Leds.rgb_on(color))
time.sleep(1)
# play_wav('people-cheering.wav')
# 중립 리액션
else:
play_wav(random.choice(neut_wavs))
leds.pattern = Pattern.blink(5)
color = (230, 0, 115)
leds.update(Leds.rgb_on(color))
time.sleep(1)
import requests
import json
from picamera import PiCamera
from time import sleep
#lib required for hearthrate emotion recognition
import RPi.GPIO as GPIO
#libs required for audio emotion recognition
import sys
import scipy.io.wavfile
sys.path.append("../lib/OpenVokaturi-3-0a/api")
print("[AUDIO] Loading library...")
import Vokaturi
Vokaturi.load("../lib/OpenVokaturi-3-0a/lib/open/pi3b.so")
print("[AUDIO] Analyzed by: %s" % Vokaturi.versionAndLicense())
import pyaudio
import wave
EMOTIONS = [
'fear', 'happy', 'neutral', 'contempt', 'surprise', 'sadness', 'anger',
'disgust'
]
_FEAR = 0
_HAPPINESS = 1
_NEUTRAL = 2
_CONTEMPT = 3
_SURPRISE = 4
_SADNESS = 5
# Paul Boersma 2017-01-25
#
# A sample script that uses the Vokaturi library to extract the emotions from
# a wav file on disk. The file has to contain a mono recording.
#
# Call syntax:
# python3 measure_wav_linux32.py path_to_sound_file.wav
import sys
import scipy.io.wavfile
sys.path.append("../api")
import Vokaturi
print ("Loading library...")
Vokaturi.load("../lib/Vokaturi_linux32.so")
print ("Analyzed by: %s" % Vokaturi.versionAndLicense())
print ("Reading sound file...")
file_name = sys.argv[1]
(sample_rate, samples) = scipy.io.wavfile.read(file_name)
print (" sample rate %.3f Hz" % sample_rate)
print ("Allocating Vokaturi sample array...")
buffer_length = len(samples)
print (" %d samples, %d channels" % (buffer_length, samples.ndim))
c_buffer = Vokaturi.SampleArrayC(buffer_length)
if samples.ndim == 1:
c_buffer[:] = samples[:] / 32768.0
else:
c_buffer[:] = 0.5*(samples[:,0]+0.0+samples[:,1]) / 32768.0
