mirror of
https://github.com/PierrunoYT/Kokoro-TTS-Local.git
synced 2025-01-27 02:30:25 +03:00
df828f0409
- Refactored the TTS generation process to initialize the model globally and load voices dynamically, improving efficiency and usability. - Introduced a new load_and_validate_voice function to ensure requested voices exist before loading, enhancing error handling. - Updated generate_tts_with_logs to provide real-time logging during speech generation, including phoneme processing and audio saving. - Improved audio conversion process with better error handling and temporary file management. - Set default voice to 'af_bella' in the Gradio interface for improved user experience.
118 lines
4.5 KiB
Python
118 lines
4.5 KiB
Python
import torch
|
|
from typing import Optional, Tuple, List
|
|
from models import build_model, load_voice, generate_speech, list_available_voices
|
|
import argparse
|
|
from tqdm.auto import tqdm
|
|
import soundfile as sf
|
|
from pathlib import Path
|
|
|
|
# Constants
|
|
SAMPLE_RATE = 22050
|
|
DEFAULT_MODEL_PATH = 'kokoro-v0_19.pth'
|
|
DEFAULT_OUTPUT_FILE = 'output.wav'
|
|
DEFAULT_LANGUAGE = 'a' # TODO: Document why this is 'a' or make configurable
|
|
DEFAULT_TEXT = "Hello, welcome to this text-to-speech test."
|
|
|
|
# Configure tqdm for better Windows console support
|
|
tqdm.monitor_interval = 0 # Disable monitor thread to prevent encoding issues
|
|
|
|
def load_and_validate_voice(voice_name: str, device: str) -> torch.Tensor:
|
|
"""Load and validate the requested voice.
|
|
|
|
Args:
|
|
voice_name: Name of the voice to load
|
|
device: Device to load the voice on ('cuda' or 'cpu')
|
|
|
|
Returns:
|
|
Loaded voice tensor
|
|
|
|
Raises:
|
|
ValueError: If the requested voice doesn't exist
|
|
"""
|
|
available_voices = list_available_voices()
|
|
if voice_name not in available_voices:
|
|
raise ValueError(f"Voice '{voice_name}' not found. Available voices: {', '.join(available_voices)}")
|
|
return load_voice(voice_name, device)
|
|
|
|
def main() -> None:
|
|
try:
|
|
# Parse command line arguments
|
|
parser = argparse.ArgumentParser(description='Kokoro TTS Demo')
|
|
parser.add_argument('--text', type=str, help='Text to synthesize (optional)')
|
|
parser.add_argument('--voice', type=str, default='af_bella', help='Voice to use (default: af_bella)')
|
|
parser.add_argument('--list-voices', action='store_true', help='List all available voices')
|
|
parser.add_argument('--model', type=str, default=DEFAULT_MODEL_PATH, help=f'Path to model file (default: {DEFAULT_MODEL_PATH})')
|
|
parser.add_argument('--output', type=str, default=DEFAULT_OUTPUT_FILE, help=f'Output WAV file (default: {DEFAULT_OUTPUT_FILE})')
|
|
parser.add_argument('--lang', type=str, default=DEFAULT_LANGUAGE, help=f'Language code (default: {DEFAULT_LANGUAGE})')
|
|
args = parser.parse_args()
|
|
|
|
if args.list_voices:
|
|
voices = list_available_voices()
|
|
print("\nAvailable voices:")
|
|
for voice in voices:
|
|
print(f"- {voice}")
|
|
return
|
|
|
|
# Set up device
|
|
device = 'cuda' if torch.cuda.is_available() else 'cpu'
|
|
print(f"Using device: {device}")
|
|
|
|
# Build model and load voice with progress indication
|
|
print("\nLoading model...")
|
|
with tqdm(total=1, desc="Building model") as pbar:
|
|
model = build_model(args.model, device)
|
|
pbar.update(1)
|
|
|
|
print("\nLoading voice...")
|
|
with tqdm(total=1, desc="Loading voice") as pbar:
|
|
try:
|
|
voice = load_and_validate_voice(args.voice, device)
|
|
pbar.update(1)
|
|
except ValueError as e:
|
|
print(f"Error: {e}")
|
|
return
|
|
|
|
# Get text input
|
|
if args.text:
|
|
text = args.text
|
|
else:
|
|
print("\nEnter the text you want to convert to speech (or press Enter for default text):")
|
|
text = input("> ").strip()
|
|
if not text:
|
|
text = DEFAULT_TEXT
|
|
|
|
print(f"\nGenerating speech for: '{text}'")
|
|
with tqdm(total=1, desc="Generating speech") as pbar:
|
|
audio, phonemes = generate_speech(model, text, voice, lang=args.lang, device=device)
|
|
pbar.update(1)
|
|
|
|
if audio is not None:
|
|
try:
|
|
if phonemes:
|
|
try:
|
|
print(f"Generated phonemes: {phonemes}")
|
|
except UnicodeEncodeError:
|
|
print("Generated phonemes: [Unicode display error - phonemes were generated but cannot be displayed]")
|
|
output_path = Path(args.output)
|
|
sf.write(output_path, audio, SAMPLE_RATE)
|
|
print(f"\nAudio saved to {output_path.absolute()}")
|
|
except Exception as e:
|
|
print(f"Error saving output: {e}")
|
|
print("Audio generation was successful, but saving failed.")
|
|
else:
|
|
print("Error: Failed to generate audio")
|
|
|
|
except Exception as e:
|
|
print(f"Error in main: {e}")
|
|
import traceback
|
|
traceback.print_exc()
|
|
finally:
|
|
# Cleanup
|
|
if 'model' in locals():
|
|
del model
|
|
if 'voice' in locals():
|
|
del voice
|
|
torch.cuda.empty_cache()
|
|
|
|
if __name__ == "__main__":
|
|
main() |