enhancement_VAE/enhance_audio_48k.py

#!/usr/bin/env python3
# coding=utf-8
# Copyright (c) 2026, NVIDIA CORPORATION.  All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""CLI launcher for Audex Enhancement VAE"""

from __future__ import annotations

import argparse
from pathlib import Path

import torch

from enhancement_vae import DEFAULT_CHECKPOINT, DEFAULT_CONFIG, enhance_file, iter_input_files, load_model


def parse_args() -> argparse.Namespace:
    script_dir = Path(__file__).resolve().parent
    parser = argparse.ArgumentParser(description="Enhance XCodec1-decoded 16 kHz WAVs to 48 kHz.")
    parser.add_argument("--input", type=Path, required=True, help="Input audio file or directory.")
    parser.add_argument("--output-dir", type=Path, required=True, help="Directory for enhanced 48 kHz WAVs.")
    parser.add_argument("--device", type=str, default="cuda" if torch.cuda.is_available() else "cpu")
    parser.add_argument("--seed", type=int, default=0, help="Torch seed for stochastic VAE sampling.")
    parser.add_argument("--deterministic", action="store_true", help="Use posterior mean instead of VAE sampling.")
    parser.set_defaults(
        checkpoint=script_dir / DEFAULT_CHECKPOINT,
        config=script_dir / DEFAULT_CONFIG,
    )
    return parser.parse_args()


def main() -> None:
    args = parse_args()
    torch.manual_seed(args.seed)
    device = torch.device(args.device)
    model = load_model(checkpoint_path=args.checkpoint, config_path=args.config, device=device)
    input_files = iter_input_files(args.input)
    if not input_files:
        raise ValueError(f"No audio files found in {args.input}")
    for input_path in input_files:
        output_path = args.output_dir / f"{input_path.stem}_enhanced_48k.wav"
        enhance_file(model, input_path, output_path, deterministic=args.deterministic)
        print(f"{input_path} -> {output_path}")


if __name__ == "__main__":
    main()