eschmidbauer / RVC_CLI

RVC CLI enables seamless interaction with Retrieval-based Voice Conversion through commands or HTTP requests.

RVC_CLI: Retrieval-based Voice Conversion Command Line Interface

Todo

• Use this style for commands python app.py infer --hop_length 128 --f0method rmvpe --input_path "input.wav" --output_path "output.wav" --pth_file "model.pth" --index_path "index_file.index"

Table of Contents

1. Installation
   • Windows
   • Linux
2. Getting Started
   • Inference
   • Training
   • Additional Features
3. API
4. Credits

Installation

Ensure you have the required Python packages installed by running (Python 3.9 is recommended):

Windows

Execute the install.bat file to activate a Conda environment. Subsequently, launch the application using env/python main.py instead of the conventional python main.py command.

Linux

chmod +x install.sh
./install.sh

Getting Started

For additional information and command-line options, refer to the help command:

python main.py -h

This command displays the available modes and their corresponding parameters, providing clarity on how to effectively use the RVC CLI.

Inference

Single Inference

python main.py infer f0up_key filter_radius index_rate hop_length f0method "input_path" "output_path" "pth_file" "index_path" split_audio

• f0up_key: Value for f0up_key (-24 to +24)
• filter_radius: Value for filter_radius (0 to 10)
• index_rate: Value for index_rate (0.0 to 1.0)
• hop_length: Value for hop_length (1 to 512)
• f0method: Value for f0method (pm, dio, crepe, crepe-tiny, harvest, rmvpe)
• input_path: Input audio path (enclosed in double quotes)
• output_path: Output audio path (enclosed in double quotes)
• pth_file: Path to the .pth file (enclosed in double quotes)
• index_path: Path to the .index file (enclosed in double quotes)
• split_audio: Value for split_audio (True or False)

Batch Inference

python main.py batch_infer f0up_key filter_radius index_rate hop_length f0method "input_folder_path" "output_folder_path" "pth_file" "index_path"

• f0up_key: Value for f0up_key (-24 to +24)
• filter_radius: Value for filter_radius (0 to 10)
• index_rate: Value for index_rate (0.0 to 1.0)
• hop_length: Value for hop_length (1 to 512)
• f0method: Value for f0method (pm, dio, crepe, crepe-tiny, harvest, rmvpe)
• input_folder_path: Input folder path (enclosed in double quotes)
• output_folder_path: Output folder path (enclosed in double quotes)
• pth_file: Path to the .pth file (enclosed in double quotes)
• index_path: Path to the .index file (enclosed in double quotes)

TTS Inference

python main.py tts_infer "tts_text" "tts_voice" f0up_key filter_radius index_rate hop_length f0method "output_tts_path" "output_rvc_path" "pth_file" "index_path"

• tts_text: Text for TTS synthesis (enclosed in double quotes)
• tts_voice: Voice for TTS synthesis (enclosed in double quotes)
• f0up_key: Value for f0up_key (-24 to +24)
• filter_radius: Value for filter_radius (0 to 10)
• index_rate: Value for index_rate (0.0 to 1.0)
• hop_length: Value for hop_length (1 to 512)
• f0method: Value for f0method (pm, dio, crepe, crepe-tiny, harvest, rmvpe)
• output_tts_path: Output TTS audio path (enclosed in double quotes)
• output_rvc_path: Output RVC audio path (enclosed in double quotes)
• pth_file: Path to the .pth file (enclosed in double quotes)
• index_path: Path to the .index file (enclosed in double quotes)

Training

Preprocess Dataset

python main.py preprocess "model_name" "dataset_path" sampling_rate

• model_name: Name of the model (enclosed in double quotes)
• dataset_path: Path to the dataset (enclosed in double quotes)
• sampling_rate: Sampling rate (32000, 40000, or 48000): Optional, default 40000

Extract Features

python main.py extract "model_name" rvc_version f0method hop_length sampling_rate

• model_name: Name of the model (enclosed in double quotes)
• rvc_version: Version of the model (v1 or v2)
• f0method: Value for f0method (pm, dio, crepe, crepe-tiny, harvest, rmvpe)
• hop_length: Value for hop_length (1 to 512)
• sampling_rate: Sampling rate (32000, 40000, or 48000)

Start Training

python main.py train "model_name" rvc_version save_every_epoch save_only_latest save_every_weights total_epoch sampling_rate batch_size gpu pitch_guidance pretrained custom_pretrained [g_pretrained] [d_pretrained]

• model_name: Name of the model (enclosed in double quotes)
• rvc_version: Version of the model (v1 or v2)
• save_every_epoch: Number of epochs after which to save the model checkpoint (1 to 50)
• save_only_latest: Save only the lastest final weight (True or False)
• save_every_weights: Save a weight every training save (True or False)
• total_epoch: Total number of training epochs (1 to 10000)
• sampling_rate: Sampling rate of the audio data (32000, 40000, or 48000): Optional, default 40000
• batch_size: Batch size, limited by GPU VRAM (4 to ∞)
• gpu: GPU number (0 to ∞ separated by -)
• pitch_guidance: Train with or without pitch guidance (True or False)
• pretrained: Train with or without pretrained models (True or False)
• custom_pretrained: Use custom pretrained models; use parameters g_/d_pretrained (True or False)
• g_pretrained_path: Path to pretrained file G, only if you have used custom_pretrained (enclosed in double quotes)
• d_pretrained_path: Path to pretrained file D, only if you have used custom_pretrained (enclosed in double quotes)

Generate Index File

python main.py index "model_name" rvc_version

• model_name: Name of the model (enclosed in double quotes)
• rvc_version: Version of the model (v1 or v2)

Additional Features

Model Information

python main.py model_information "pth_path"

• pth_path: Path to the .pth file (enclosed in double quotes)

Model Fusion

python main.py model_fusion "model_name" "pth_path_1" "pth_path_2"

• model_name: Name of the model (enclosed in double quotes)
• pth_path_1: Path to the first .pth file (enclosed in double quotes)
• pth_path_2: Path to the second .pth file (enclosed in double quotes)

Launch TensorBoard

python main.py tensorboard

Download Models

Run the download script with the following command:

python main.py download "model_link"

• model_link: Link of the model (enclosed in double quotes; Google Drive or Hugging Face)

API

To use the RVC CLI via the API, you can utilize the provided api.py script. Make API requests to the following endpoints:

• Infer: /infer
• Batch Infer: /batch_infer
• TTS: /tts
• Preprocess: /preprocess
• Extract: /extract
• Train: /train
• Index: /index
• Model Information: /model_information
• Model Fusion: /model_fusion
• Tensorboard: /tensorboard
• Download: /download

You can make POST requests to these endpoints with the required parameters in the JSON format.

Example using curl for Infer:

curl -X POST -H "Content-Type: application/json" -d '{"f0up_key": 0, "filter_radius": 5, "index_rate": 0.5, "hop_length": 256, "f0method": "rmvpe", "input_path": "input.wav", "output_path": "output.wav", "pth_file": "model.pth", "index_path": "index_file.index", "split_audio": false}' http://localhost:5000/infer

Credits

The RVC CLI is built on the foundations of the following projects:

• Mangio-RVC-Fork
• Retrieval-based-Voice-Conversion-WebUI

We acknowledge and appreciate the contributions of the respective authors and communities involved in these projects.