-
Notifications
You must be signed in to change notification settings - Fork 1
/
Speech_Generator.py
230 lines (179 loc) · 7.38 KB
/
Speech_Generator.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
import argparse
import csv
import json
import numpy as np
import os
from pydub import AudioSegment
import sys
import time
import wave
from google.cloud import texttospeech
disallowed_set = {
"PATH",
"INTENT",
"",
"GENERIC",
"GENNAME"
}
class Audio_Data_Gen:
def __init__(self, credentials:str="auth.json",
accent:str="US",
speaker:str="D") -> None:
self.accent = accent
self.speaker = speaker
data_dir_name = "response_data"
# get path for credentials
self.credential_path = credentials
os.environ['GOOGLE_APPLICATION_CREDENTIALS'] = self.credential_path
# instantiate GCP TTS objects
self.tts_client = texttospeech.TextToSpeechClient()
# contains the chunks of streamed data
self.frames = []
# Path to the response data
self.data_dir_path = os.path.join(os.getcwd(), data_dir_name)
# Creates data path if DNE
if (not(os.path.isdir(self.data_dir_path))):
os.makedirs(self.data_dir_path)
self.file_mapping = {
"<<GENERICS>>":set()
}
def parse_rows(self, paths:list) -> set:
'''
Takes multiple CSVs, reads each row-by-row, and stores the answers
as a dict. Updates the file mapping dict on generic data
An example is as follows:
answers_set = {
"Sorry, I do not know the answer to your question.",
"The ranch is 3200 acres.",
"Fred Swanton"
}
:param paths: List of paths to the data CSVs
:returns answers_set: Answer string within the set
'''
answers_set = {
"Sorry, I do not know the answer to your question."
}
for path in paths:
rows = self.read_csv(path)
generic_name = ""
for row in rows:
# Extract the cell from the 2nd column
answer = row[1]
if (row[0] not in disallowed_set):
# Add to set if it is not in the set
if (answer not in answers_set):
answers_set.add(answer)
elif (row[0] == "GENNAME"):
generic_name = row[1]
self.file_mapping["<<GEN>>%s" % generic_name] = []
self.file_mapping["<<GENERICS>>"].add(generic_name)
elif (row[0] == "GENERIC"):
if (answer not in answers_set):
answers_set.add(answer)
self.file_mapping["<<GEN>>%s" % generic_name].append(answer)
return answers_set
def read_csv(self, csv_path: str) -> list:
'''
Reads the CSV row-by-row and stores the rows as a list of strs.
:param csv_path: Path to the QA pairs CSV
:returns rows: List of strs of each row of a CSV
'''
rows = []
# Open the CSV
with open(csv_path, 'r') as csv_file:
# Instantiate CSV object based on input CSV
csvreader = csv.reader(csv_file)
# Remove the CSV header
fields = next(csvreader)
# Iterate through each row
for row in csvreader:
rows.append(row)
return rows
def Text_To_Speech(self, csv_paths:list) -> None:
'''
Iterates through the set of answer strings and generates audio of the
string which is stored with a corresponding JSON file to map audio
files to the answer string.
:param csv_path: Paths to the QA pairs CSVs
:returns N/A
'''
gen_name = ""
sample_count = 0
dataset = self.parse_rows(csv_paths)
for answer_str in dataset:
if answer_str not in self.file_mapping["<<GENERICS>>"]:
# Build the voice request, select the language code
# ("en-US") and the ssml voice gender ("neutral")
voice = texttospeech.VoiceSelectionParams(
language_code='en-%s' % self.accent,
name='en-%s-WaveNet-%s' % (self.accent, self.speaker))
# Select the type of audio file you want returned
audio_config = texttospeech.AudioConfig(
audio_encoding=texttospeech.AudioEncoding.LINEAR16,
sample_rate_hertz=16000)
# Set the text input to be synthesized
synthesis_input = texttospeech.SynthesisInput(text=answer_str)
# Perform the text-to-speech request on the text input with
# the selected voice parameters and audio file type
answer_audio = self.tts_client.synthesize_speech(
input=synthesis_input,
voice=voice,
audio_config=audio_config)
self.store_data(
answer_str,
answer_audio,
sample_count)
sample_count += 1
def store_data(self,
answer_str:str,
answer_audio:list,
sample_count:int) -> None:
'''
Takes the answer str, the audio of the str, and the current file number
and stores the audio & JSON
:param answer_str: string containing the current answer
:param answer_audio: audio byte array of the synthesized string
:param sample_count: integer count of the current file #
'''
file_name = "response_%s.wav" % str(sample_count)
file_path = os.path.join(self.data_dir_path, file_name)
with open(file_path, 'wb') as out:
out.write(answer_audio.audio_content)
sound = AudioSegment.from_wav(file_path)
sound = sound.set_channels(1)
sound.export(file_path, format="wav")
print("Audio String: %s => File Written: %s" % (answer_str, file_name))
self.file_mapping[answer_str] = file_name
self.file_mapping["<<GENERICS>>"] = list(self.file_mapping["<<GENERICS>>"])
json_path = os.path.join(self.data_dir_path, "answer_to_file.json")
with open(json_path, "w") as outfile:
json.dump(self.file_mapping, outfile, indent=3)
if __name__ == "__main__":
parser = argparse.ArgumentParser(
description='Generates audio responses for the answers of the\
QA pairs CSV')
parser.add_argument(
'--json',
dest="json",
help='Authentication JSON',
required=True)
parser.add_argument(
'--csv',
dest="csv",
help='QA Pairs CSV',
nargs="+",
type=str,
required=True)
parser.add_argument('--accent',
dest="accent",
help='Accent of the speaker',
default="US")
parser.add_argument('--speaker',
dest="speaker",
help='Speaker voice (US: A - F | UK: A - F | AU: A - D)',
default="D")
args = parser.parse_args()
data_gen = Audio_Data_Gen(args.json,
args.accent,
args.speaker)
data_gen.Text_To_Speech(args.csv)