Spaces:
Sleeping
Sleeping
File size: 1,297 Bytes
5efbc82 9acb9c3 9a88d9c 9acb9c3 5efbc82 9a88d9c 9acb9c3 |
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 |
import os
# Configure cache directories for Hugging Face Spaces
os.environ['HF_HOME'] = '/tmp/hf_cache'
os.environ['TRANSFORMERS_CACHE'] = '/tmp/hf_cache'
os.environ['HF_HUB_CACHE'] = '/tmp/hf_cache'
os.environ['TORCH_HOME'] = '/tmp/torch_cache'
# Create cache directories
os.makedirs('/tmp/hf_cache', exist_ok=True)
os.makedirs('/tmp/torch_cache', exist_ok=True)
from kokoro import KPipeline
import soundfile as sf
import torch
# Initialize Kokoro pipeline
pipeline = KPipeline(lang_code='a', repo_id='hexgrad/Kokoro-82M')
# Text to convert to speech
text = '''
[Kokoro](/kˈOkəɹO/) is an open-weight TTS model with 82 million parameters. Despite its lightweight architecture, it delivers comparable quality to larger models while being significantly faster and more cost-efficient. With Apache-licensed weights, [Kokoro](/kˈOkəɹO/) can be deployed anywhere from production environments to personal projects.
'''
# Generate speech using Kokoro
generator = pipeline(text, voice='af_heart')
# Process and save the generated audio
for i, (gs, ps, audio) in enumerate(generator):
print(f"Segment {i}: gs={gs}, ps={ps}")
# Save each segment as a separate file
sf.write(f'{i}.wav', audio, 24000)
print(f"Saved segment {i} as {i}.wav")
print("Speech generation completed!")
|