225 lines
6.7 KiB
Python
225 lines
6.7 KiB
Python
#!/usr/bin/env python3
|
|
"""
|
|
F5-TTS Integration Guide for ChatterBox Voice
|
|
Complete setup and usage instructions for F5-TTS integration
|
|
"""
|
|
|
|
# ===== INSTALLATION GUIDE =====
|
|
|
|
"""
|
|
F5-TTS Installation Steps:
|
|
|
|
1. Install F5-TTS from source (recommended):
|
|
```bash
|
|
git clone https://github.com/SWivid/F5-TTS.git
|
|
cd F5-TTS
|
|
pip install -e .
|
|
```
|
|
|
|
2. Install F5-TTS specific dependencies:
|
|
```bash
|
|
pip install -r requirements_f5tts.txt
|
|
```
|
|
|
|
3. Optional: Place F5-TTS models in ComfyUI models directory:
|
|
```
|
|
ComfyUI/models/F5-TTS/
|
|
├── F5TTS_Base/
|
|
│ ├── model_1200000.safetensors
|
|
│ └── vocab.txt
|
|
├── F5TTS_v1_Base/
|
|
│ ├── model_1250000.safetensors
|
|
│ └── vocab.txt
|
|
└── E2TTS_Base/
|
|
├── model_1200000.safetensors
|
|
└── vocab.txt
|
|
```
|
|
|
|
4. Restart ComfyUI to load F5-TTS nodes
|
|
"""
|
|
|
|
# ===== USAGE EXAMPLES =====
|
|
|
|
# Basic F5-TTS Usage
|
|
f5tts_basic_example = {
|
|
"text": "Hello! This is F5-TTS integrated with ChatterBox Voice. It provides high-quality text-to-speech with voice cloning capabilities.",
|
|
"ref_text": "This is the reference text that matches the reference audio for voice cloning.",
|
|
"reference_audio": "<audio_input>", # ComfyUI audio input
|
|
"model": "F5TTS_Base",
|
|
"device": "auto",
|
|
"temperature": 0.8,
|
|
"speed": 1.0,
|
|
"target_rms": 0.1,
|
|
"nfe_step": 32,
|
|
"cfg_strength": 2.0,
|
|
"enable_chunking": True,
|
|
"max_chars_per_chunk": 400,
|
|
"chunk_combination_method": "auto"
|
|
}
|
|
|
|
# Long Text Processing with Chunking
|
|
f5tts_long_text_example = {
|
|
"text": """This is a very long text that will be automatically chunked by the F5-TTS system.
|
|
The chunking system intelligently splits the text while preserving sentence boundaries and
|
|
ensuring natural flow. Each chunk is processed separately and then combined using the
|
|
specified method to create seamless audio output.""",
|
|
"ref_text": "This is my reference voice sample.",
|
|
"reference_audio": "<audio_input>",
|
|
"model": "F5TTS_Base",
|
|
"enable_chunking": True,
|
|
"max_chars_per_chunk": 300,
|
|
"chunk_combination_method": "crossfade",
|
|
"silence_between_chunks_ms": 100
|
|
}
|
|
|
|
# Multi-language Support
|
|
f5tts_multilingual_examples = {
|
|
"german": {
|
|
"text": "Hallo! Dies ist F5-TTS mit deutscher Sprachunterstützung.",
|
|
"ref_text": "Dies ist meine deutsche Referenzstimme.",
|
|
"model": "F5-DE"
|
|
},
|
|
"spanish": {
|
|
"text": "¡Hola! Este es F5-TTS con soporte para español.",
|
|
"ref_text": "Esta es mi voz de referencia en español.",
|
|
"model": "F5-ES"
|
|
},
|
|
"french": {
|
|
"text": "Bonjour! Ceci est F5-TTS avec support français.",
|
|
"ref_text": "Ceci est ma voix de référence française.",
|
|
"model": "F5-FR"
|
|
},
|
|
"japanese": {
|
|
"text": "こんにちは！これは日本語対応のF5-TTSです。",
|
|
"ref_text": "これは日本語のリファレンス音声です。",
|
|
"model": "F5-JP"
|
|
}
|
|
}
|
|
|
|
# ===== INTEGRATION CODE SNIPPETS =====
|
|
|
|
# For developers extending the F5-TTS integration:
|
|
|
|
def example_f5tts_usage():
|
|
"""Example of using F5-TTS in custom code"""
|
|
from chatterbox.f5tts import ChatterBoxF5TTS
|
|
|
|
# Initialize F5-TTS
|
|
f5tts = ChatterBoxF5TTS.from_pretrained("cuda", "F5TTS_Base")
|
|
|
|
# Generate speech
|
|
audio = f5tts.generate(
|
|
text="Hello world!",
|
|
ref_audio_path="reference.wav",
|
|
ref_text="This is reference text",
|
|
temperature=0.8,
|
|
speed=1.0
|
|
)
|
|
|
|
return audio
|
|
|
|
def example_model_discovery():
|
|
"""Example of discovering available F5-TTS models"""
|
|
from chatterbox.f5tts.f5tts import find_f5tts_models, get_f5tts_models
|
|
|
|
# Find model paths
|
|
model_paths = find_f5tts_models()
|
|
print("Available model paths:", model_paths)
|
|
|
|
# Get model names
|
|
model_names = get_f5tts_models()
|
|
print("Available models:", model_names)
|
|
|
|
def example_error_handling():
|
|
"""Example of proper error handling with F5-TTS"""
|
|
try:
|
|
from chatterbox.f5tts import ChatterBoxF5TTS, F5TTS_AVAILABLE
|
|
|
|
if not F5TTS_AVAILABLE:
|
|
print("F5-TTS not available - please install dependencies")
|
|
return None
|
|
|
|
f5tts = ChatterBoxF5TTS.from_pretrained("cuda", "F5TTS_Base")
|
|
|
|
# Your F5-TTS code here
|
|
|
|
except ImportError as e:
|
|
print(f"F5-TTS import error: {e}")
|
|
return None
|
|
except Exception as e:
|
|
print(f"F5-TTS error: {e}")
|
|
return None
|
|
|
|
# ===== TROUBLESHOOTING =====
|
|
|
|
"""
|
|
Common Issues and Solutions:
|
|
|
|
1. F5-TTS not available:
|
|
- Install F5-TTS from source: git clone https://github.com/SWivid/F5-TTS.git
|
|
- Install dependencies: pip install -r requirements_f5tts.txt
|
|
|
|
2. Model loading errors:
|
|
- Check model files are in correct format (.safetensors or .pt)
|
|
- Verify vocab.txt file exists alongside model file
|
|
- Try different model variants if one fails
|
|
|
|
3. Audio quality issues:
|
|
- Ensure reference audio is high quality (24kHz recommended)
|
|
- Reference text should accurately match reference audio
|
|
- Adjust temperature and cfg_strength parameters
|
|
|
|
4. Memory issues:
|
|
- Enable chunking for long texts
|
|
- Reduce max_chars_per_chunk
|
|
- Use CPU instead of GPU if needed
|
|
|
|
5. Performance optimization:
|
|
- Use local models instead of downloading
|
|
- Enable GPU acceleration if available
|
|
- Use appropriate chunk sizes for your hardware
|
|
"""
|
|
|
|
# ===== CONFIGURATION =====
|
|
|
|
# F5-TTS Configuration Options
|
|
F5TTS_CONFIG = {
|
|
# Model Selection
|
|
"available_models": [
|
|
"F5TTS_Base", # English base model
|
|
"F5TTS_v1_Base", # English v1 model
|
|
"E2TTS_Base", # E2-TTS model
|
|
"F5-DE", # German model
|
|
"F5-ES", # Spanish model
|
|
"F5-FR", # French model
|
|
"F5-JP", # Japanese model
|
|
"F5-IT", # Italian model
|
|
"F5-TH", # Thai model
|
|
],
|
|
|
|
# Audio Settings
|
|
"sample_rate": 24000,
|
|
"audio_format": "float32",
|
|
|
|
# Generation Parameters
|
|
"default_temperature": 0.8,
|
|
"default_speed": 1.0,
|
|
"default_target_rms": 0.1,
|
|
"default_nfe_step": 32,
|
|
"default_cfg_strength": 2.0,
|
|
|
|
# Chunking Settings
|
|
"default_chunk_size": 400,
|
|
"chunk_methods": ["auto", "concatenate", "silence_padding", "crossfade"],
|
|
|
|
# Model Paths
|
|
"comfyui_models_dir": "ComfyUI/models/F5-TTS/",
|
|
"fallback_to_huggingface": True,
|
|
}
|
|
|
|
if __name__ == "__main__":
|
|
print("F5-TTS Integration Guide")
|
|
print("=" * 50)
|
|
print("This guide contains setup instructions and usage examples for F5-TTS integration.")
|
|
print("Please read the comments in this file for detailed information.")
|
|
print("=" * 50) |