Spaces:
Running
Running
| # ================================================================================================== | |
| # DEEPFAKE AUDIO - synthesizer_train.py (Acoustic Generator Training) | |
| # ================================================================================================== | |
| # | |
| # 📝 DESCRIPTION | |
| # This script manages the training of the Tacotron 2-based Synthesizer. It optimizes | |
| # a sequence-to-sequence neural network with attention to map text tokens into | |
| # realistic mel-spectrograms. By conditioning the process on speaker embeddings, | |
| # the model learns to synthesize speech in the voice of any target speaker. | |
| # | |
| # 👤 AUTHORS | |
| # - Amey Thakur (https://github.com/Amey-Thakur) | |
| # - Mega Satish (https://github.com/msatmod) | |
| # | |
| # 🤝🏻 CREDITS | |
| # Original Real-Time Voice Cloning methodology by CorentinJ | |
| # Repository: https://github.com/CorentinJ/Real-Time-Voice-Cloning | |
| # | |
| # 🔗 PROJECT LINKS | |
| # Repository: https://github.com/Amey-Thakur/DEEPFAKE-AUDIO | |
| # Video Demo: https://youtu.be/i3wnBcbHDbs | |
| # Research: https://github.com/Amey-Thakur/DEEPFAKE-AUDIO/blob/main/DEEPFAKE-AUDIO.ipynb | |
| # | |
| # 📜 LICENSE | |
| # Released under the MIT License | |
| # Release Date: 2021-02-06 | |
| # ================================================================================================== | |
| from pathlib import Path | |
| from synthesizer.hparams import hparams | |
| from synthesizer.train import train | |
| from utils.argutils import print_args | |
| import argparse | |
| if __name__ == "__main__": | |
| # --- TRAINING PARAMETERS --- | |
| parser = argparse.ArgumentParser( | |
| description="Synthesizer Training Hub: Optimizing the Tacotron 2 sequence-to-sequence engine.", | |
| formatter_class=argparse.ArgumentDefaultsHelpFormatter | |
| ) | |
| # --- SESSION DEFINITION --- | |
| parser.add_argument("run_id", type=str, | |
| help="Identifier for this training experiment. Logs and weights will be archived here.") | |
| parser.add_argument("syn_dir", type=Path, | |
| help="Path to the training data directory (spectrograms, wavs, and embeds).") | |
| # --- STORAGE & BACKUP --- | |
| parser.add_argument("-m", "--models_dir", type=Path, default="saved_models", | |
| help="Destination for serialized model weights and event logs.") | |
| parser.add_argument("-s", "--save_every", type=int, default=1000, | |
| help="Iteration interval for materializing weights onto long-term storage.") | |
| parser.add_argument("-b", "--backup_every", type=int, default=25000, | |
| help="Frequency of immutable state backups to prevent data loss.") | |
| parser.add_argument("-f", "--force_restart", action="store_true", | |
| help="Inhibit weight loading and re-initialize from scratch.") | |
| # --- HYPERPARAMETER TUNING --- | |
| parser.add_argument("--hparams", default="", | |
| help="Sequence architecture overrides (comma-separated).") | |
| args = parser.parse_args() | |
| # --- EXECUTION --- | |
| print_args(args, parser) | |
| print("🤝🏻 Scholarly Partnership: Amey Thakur & Mega Satish") | |
| print("🚀 Initiating Synthetic Engine Training - Optimizing attention alignments...") | |
| args.hparams = hparams.parse(args.hparams) | |
| # Delegate to the sequence-to-sequence training engine. | |
| train(**vars(args)) | |