RVC Text-to-Speech

RVC Text-to-Speech (rvc-tts) is a Python package that integrates RVC (Retrieval-based Voice Conversion) models with edge-tts to provide advanced text-to-speech capabilities.

Features

• Text-to-Speech Conversion: Converts input text into speech using Microsoft's neural voices via edge-tts.
• Voice Conversion: Transforms the generated speech into a target voice using RVC models.
• Customization Options: Allows adjustments to speech speed, pitch, and other parameters for personalized output.

Installation

Ensure you have Python 3.8 or higher installed. You can install the rvc-tts package using pip:

pip install rvc-tts

Usage

Here's an example of how to use the rvc-tts package:

from rvc_tts import RVCTTS

# Initialize the TTS engine
tts = RVCTTS()

# Generate speech
info, edge_output_filename, audio_opt = tts.tts(
    model_name="model1",               # Name of the RVC model to use
    speed=0,                           # Speech speed adjustment (0 for default)
    tts_text="Hello, world!",          # Text to convert to speech
    tts_voice="en-US-JennyNeural",     # Voice selection from edge-tts
    f0_up_key=0,                       # Pitch adjustment
    f0_method="rmvpe",                 # Pitch extraction method
    index_rate=1,                      # Index rate for voice conversion
    protect=0                          # Protection parameter (refer to RVC documentation)
)

Parameters

• model_name: Specifies the RVC model to use for voice conversion.
• speed: Adjusts the speech speed. Positive values increase speed, negative values decrease it, and 0 maintains the default speed.
• tts_text: The input text string that you want to convert to speech.
• tts_voice: Selects the voice for speech synthesis. A list of available voices can be found in the edge-tts documentation.
• f0_up_key: Adjusts the pitch of the output speech. Positive values raise the pitch, negative values lower it, and 0 leaves it unchanged.
• f0_method: Specifies the pitch extraction method. Options include "rmvpe" and others as detailed in the RVC documentation.
• index_rate: Controls the blending rate between the original and converted voice. A value of 1 uses only the converted voice, while lower values mix in more of the original.
• protect: A parameter for advanced users to control specific aspects of voice conversion. Refer to the RVC documentation for detailed usage.

Requirements

• Python: Version 3.8 or higher.
• PyTorch: Ensure that PyTorch is installed in your environment. Installation instructions can be found on the official PyTorch website.
• ffmpeg: Required for audio processing. Installation instructions are available on the ffmpeg official website.

Additional Resources

• RVC Documentation: For more details on RVC models and their capabilities, visit the RVC GitHub repository.
• edge-tts Documentation: For information on available voices and additional settings, refer to the edge-tts GitHub repository.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Acknowledgments

Special thanks to the developers of RVC and edge-tts for their contributions to the open-source community.