Spaces:
Sleeping
Sleeping
File size: 10,873 Bytes
aa7cb02 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 |
import pandas as pd
import numpy as np
import os
import sys
from models.parakeet import parakeet_ctc_model, parakeet_ctc_process
from models.es_fastconformer import stt_es_model, stt_es_process
from pydub import AudioSegment
import json
class Data_Pipeline():
"""
A class to handle data processing and transcription using various models.
Args:
wav_dir (str): Directory containing the .wav files.
Example: "path/to/wav_files/"
lang (str): Language of the audio files, either 'en' for English or 'es' for Spanish.
Example: "en"
"""
def __init__(self, wav_dir, lang):
self.wav_dir = wav_dir
self.lang = lang
def get_wav_files(self):
"""
Retrieve all .wav files from the specified directory.
Returns:
list: A list of paths to the .wav files.
Example: ["path/to/file1.wav", "path/to/file2.wav"]
"""
self.wav_files = []
for root, dirs, files in os.walk(self.wav_dir):
for file in files:
if file.endswith('.wav'):
# make the path into the full path to the file
file = os.path.join(self.wav_dir, file)
self.wav_files.append(file)
return self.wav_files
def get_combined_wav_lengths(self):
"""
Calculate the total length of all .wav files in the directory.
Returns:
float: The total length of all .wav files in seconds.
Example: 123.45
"""
# returns a float number of the total length of all the wav files in the directory
total_length = 0
for file in self.wav_files:
if file.endswith('.wav'):
audio = AudioSegment.from_file(file)
total_length += len(audio)
return total_length/1000
def estimated_transcription_time(self):
"""
Estimate the transcription time based on the real-time factor.
Returns:
float: The estimated transcription time in seconds.
Example: 185.175
str: Error message if the language is not supported.
Example: "Error: Language not supported."
"""
# returns a float number of the estimated time based on the real time factor
# real time factor is 1.5
if self.lang == "en":
return self.get_combined_wav_lengths() * 1.5
elif self.lang == "es":
return self.get_combined_wav_lengths() * 1.5
return "Error: Language not supported."
def load_models(self):
"""
Load the appropriate ASR model based on the language.
Returns:
object: The loaded ASR model.
Example: <nemo.collections.asr.models.EncDecCTCModelBPE object>
str: Error message if the language is not supported.
Example: "Error: Language not supported."
"""
if self.lang == "en":
self.parakeet_model = parakeet_ctc_model()
return self.parakeet_model
elif self.lang == "es":
self.es_model = stt_es_model()
return self.es_model
return "Error: Language not supported."
def en_transcribe(self, audio_file):
"""
Transcribe an English audio file using the Parakeet CTC model.
Args:
audio_file (str): Path to the audio file.
Example: "path/to/audio_file.wav"
Returns:
list: A list containing the transcribed text.
Example: ["transcribed text"]
"""
text = parakeet_ctc_process(self.parakeet_model, audio_file)
return text
def es_transcribe(self, audio_file):
"""
Transcribe a Spanish audio file using the FastConformer model.
Args:
audio_file (str): Path to the audio file.
Example: "path/to/audio_file.wav"
Returns:
list: A list containing the transcribed text.
Example: ["transcribed text"]
"""
text = stt_es_process(self.es_model, audio_file)
return text
def read_transcriptions(self, json_path):
"""
Read transcriptions from a JSON file.
Args:
json_path (str): Path to the JSON file.
Example: "path/to/data.json"
Returns:
dict: The data read from the JSON file.
Example: {"text": ["text1", "text2"], "original_path": ["path1", "path2"]}
"""
# read the json file
with open(json_path) as f:
self.data = json.load(f)
return self.data
def get_transcription(self, file_path):
"""
Get the transcription for a specific file from the JSON data.
Args:
file_path (str): Path to the original audio file.
Example: "path/to/audio_file.wav"
Returns:
str: The transcription for the specified file.
Example: "This is the transcription."
str: Error message if no transcription is found.
Example: "Error: No transcription found."
"""
# the json file has the following keys: text, original_path, path_to_save, language, order, original_text
# get the "original_text" of the element that has the "original_path" equal to the file_path
for i in range(len(self.data['original_path'])):
if self.data['original_path'][i] == file_path:
return self.data['original_text'][i]
return "Error: No transcription found."
def data_formatter_with_models(self):
"""
Format data by transcribing audio files using the appropriate models.
Returns:
pd.DataFrame: A DataFrame containing the transcriptions.
Example: pd.DataFrame({'wav_file': ["file1", "file2"], 'transcription': ["text1", "text2"], 'transcription2': ["text1", "text2"], 'speaker_name': ["user0", "user0"]})
str: Error message if the language is not supported.
Example: "Error: Language not supported."
"""
self.transcriptions_df = pd.DataFrame(columns = ['wav_file', 'transcription','transcription2' ])
if self.lang == "en":
self.load_models()
self.get_wav_files()
for file in self.wav_files:
if file.endswith('.wav'):
transcription = parakeet_ctc_process(self.parakeet_model, file)
# append transcriptions_df with the wav_file and transcription
self.transcriptions_df = self.transcriptions_df.append({'wav_file': file, 'transcription': transcription[0], 'transcription2': transcription[0],'speaker_name': "user0"}, ignore_index=True)
return self.transcriptions_df
elif self.lang == "es":
self.load_models()
self.get_wav_files()
for file in self.wav_files:
if file.endswith('.wav'):
self.transcriptions_df = stt_es_process(self.es_model, file)
# make the path into the full path to the file
file = os.path.join(self.wav_dir, file)
# append transcriptions_df with the wav_file and transcription
self.transcriptions_df = self.transcriptions_df.append({'wav_file': file, 'transcription': transcription[0], 'transcription2': transcription[0], 'speaker_name':"user0"}, ignore_index=True)
return self.transcriptions_df
return "Error: Language not supported."
def data_formatter_without_models(self):
"""
Format data by retrieving transcriptions from a JSON file and transcribing any missing data.
Returns:
pd.DataFrame: A DataFrame containing the transcriptions.
Example: pd.DataFrame({'wav_file': ["file1", "file2"], 'transcription': ["text1", "text2"], 'transcription2': ["text1", "text2"], 'speaker_name': ["user0", "user0"]})
"""
self.transcriptions_df = pd.DataFrame(columns = ['wav_file', 'transcription','transcription2' ])
self.get_wav_files()
for file in self.wav_files:
if file.endswith('.wav'):
transcription = self.get_transcription(file)
if transcription == "Error: No transcription found." and self.lang == "en":
transcription = parakeet_ctc_process(self.parakeet_model, file)
elif transcription == "Error: No transcription found." and self.lang == "es":
transcription = stt_es_process(self.es_model, file)
# make the path into the full path to the file
#file = os.path.join(self.wav_dir, file)
# append transcriptions_df with the wav_file and transcription
self.transcriptions_df = self.transcriptions_df.append({'wav_file': file, 'transcription': transcription[0], 'transcription2': transcription[0], 'speaker_name': "user0"}, ignore_index=True)
return self.transcriptions_df
def save_transcriptions(self, output_file):
"""
Save the transcriptions to CSV files, splitting into training and evaluation datasets.
Args:
output_file (str): Base path for the output CSV files.
Example: "path/to/output"
Returns:
tuple: A tuple containing a success message and paths to the training and evaluation CSV files.
Example: ("Data saved successfully.", "path/to/output_train.csv", "path/to/output_eval.csv")
"""
# split the data into two data, train and eval data
from sklearn.model_selection import train_test_split
train_data, eval_data = train_test_split(self.transcriptions_df, test_size=0.2, random_state=42)
# save the data into csv files
self.path_to_train_data = output_file + "_train.csv"
self.path_to_eval_data = output_file + "_eval.csv"
train_data.to_csv(output_file + "_train.csv" , index=False, sep='|')
eval_data.to_csv(output_file + "_eval.csv", index=False, sep='|')
return "Data saved successfully.", self.path_to_train_data, self.path_to_eval_data
def get_paths(self):
"""
Retrieve the paths to the training and evaluation CSV files and the .wav files directory.
Returns:
tuple: A tuple containing paths to the training data, evaluation data, and .wav files directory.
Example: ("path/to/train.csv", "path/to/eval.csv", "path/to/wav_files/")
"""
# csv files, wav files directory
return self.path_to_train_data, self.path_to_eval_data, self.wav_dir
|