mirror of
https://github.com/storytold/storyteller-ml.git
synced 2026-10-13 23:53:02 +00:00
michael chung
48 lines
1.7 KiB
Python
48 lines
1.7 KiB
Python
from phonemizer import phonemize
|
|
from phonemizer.separator import Separator
|
|
import os
|
|
|
|
def phonemize_files(file_paths, language='en-us', backend='espeak', output_prefix='phonemized_'):
|
|
"""
|
|
Phonemizes the text in each of the specified files and writes the results to new files.
|
|
|
|
Parameters:
|
|
- file_paths: A list of strings, where each string is a path to a file to be processed.
|
|
- language: The language code to use for phonemization.
|
|
- backend: The phonemization backend to use.
|
|
- output_prefix: Prefix for the output file names.
|
|
|
|
Returns:
|
|
A list of paths to the output files containing the phonemized text.
|
|
"""
|
|
output_files = []
|
|
|
|
def phonemize_text(text):
|
|
return phonemize(
|
|
text,
|
|
language=language,
|
|
backend=backend,
|
|
preserve_punctuation=True,
|
|
strip=True)
|
|
|
|
for input_file in file_paths:
|
|
base_name = os.path.basename(input_file)
|
|
output_file = f'{output_prefix}{base_name}'
|
|
output_files.append(output_file)
|
|
|
|
with open(input_file, 'r', encoding='utf-8') as infile, open(output_file, 'w', encoding='utf-8') as outfile:
|
|
for line in infile:
|
|
path_to_wav, text, speaker_id = line.strip().split('|')
|
|
phonemized_text = phonemize_text(text)
|
|
outfile.write(f'{path_to_wav}|{phonemized_text}|{speaker_id}\n')
|
|
|
|
return output_files
|
|
|
|
|
|
def main():
|
|
file_paths = ['Data/OOD_harness.txt', 'Data/train_harness.txt','Data/val_harness.txt'] # Replace these with your actual file paths
|
|
output_files = phonemize_files(file_paths)
|
|
print("Output files:", output_files)
|
|
|
|
if __name__ == "__main__":
|
|
main() |