infer_uvr5.py

import os, sys, torch, warnings, pdb

warnings.filterwarnings("ignore")
import librosa
import importlib
import numpy as np
import hashlib, math
from tqdm import tqdm
from uvr5_pack.lib_v5 import spec_utils
from uvr5_pack.utils import _get_name_params, inference
from uvr5_pack.lib_v5.model_param_init import ModelParameters
from scipy.io import wavfile


class _audio_pre_:
    def __init__(self, agg, model_path, device, is_half):
        self.model_path = model_path
        self.device = device
        self.data = {
            # Processing Options
            "postprocess": False,
            "tta": False,
            # Constants
            "window_size": 512,
            "agg": agg,
            "high_end_process": "mirroring",
        }
        nn_arch_sizes = [
            31191,  # default
            33966,
            61968,
            123821,
            123812,
            537238,  # custom
        ]
        self.nn_architecture = list("{}KB".format(s) for s in nn_arch_sizes)
        model_size = math.ceil(os.stat(model_path).st_size / 1024)
        nn_architecture = "{}KB".format(
            min(nn_arch_sizes, key=lambda x: abs(x - model_size))
        )
        nets = importlib.import_module(
            "uvr5_pack.lib_v5.nets"
            + f"_{nn_architecture}".replace("_{}KB".format(nn_arch_sizes[0]), ""),
            package=None,
        )
        model_hash = hashlib.md5(open(model_path, "rb").read()).hexdigest()
        param_name, model_params_d = _get_name_params(model_path, model_hash)

        mp = ModelParameters(model_params_d)
        model = nets.CascadedASPPNet(mp.param["bins"] * 2)
        cpk = torch.load(model_path, map_location="cpu")
        model.load_state_dict(cpk)
        model.eval()
        if is_half:
            model = model.half().to(device)
        else:
            model = model.to(device)

        self.mp = mp
        self.model = model

    def _path_audio_(self, music_file, ins_root=None, vocal_root=None):
        if ins_root is None and vocal_root is None:
            return "No save root."
        name = os.path.basename(music_file)
        if ins_root is not None:
            os.makedirs(ins_root, exist_ok=True)
        if vocal_root is not None:
            os.makedirs(vocal_root, exist_ok=True)
        X_wave, y_wave, X_spec_s, y_spec_s = {}, {}, {}, {}
        bands_n = len(self.mp.param["band"])
        # print(bands_n)
        for d in range(bands_n, 0, -1):
            bp = self.mp.param["band"][d]
            if d == bands_n:  # high-end band
                (
                    X_wave[d],
                    _,
                ) = librosa.core.load(  # 理论上librosa读取可能对某些音频有bug，应该上ffmpeg读取，但是太麻烦了弃坑
                    music_file,
                    bp["sr"],
                    False,
                    dtype=np.float32,
                    res_type=bp["res_type"],
                )
                if X_wave[d].ndim == 1:
                    X_wave[d] = np.asfortranarray([X_wave[d], X_wave[d]])
            else:  # lower bands
                X_wave[d] = librosa.core.resample(
                    X_wave[d + 1],
                    self.mp.param["band"][d + 1]["sr"],
                    bp["sr"],
                    res_type=bp["res_type"],
                )
            # Stft of wave source
            X_spec_s[d] = spec_utils.wave_to_spectrogram_mt(
                X_wave[d],
                bp["hl"],
                bp["n_fft"],
                self.mp.param["mid_side"],
                self.mp.param["mid_side_b2"],
                self.mp.param["reverse"],
            )
            # pdb.set_trace()
            if d == bands_n and self.data["high_end_process"] != "none":
                input_high_end_h = (bp["n_fft"] // 2 - bp["crop_stop"]) + (
                    self.mp.param["pre_filter_stop"] - self.mp.param["pre_filter_start"]
                )
                input_high_end = X_spec_s[d][
                    :, bp["n_fft"] // 2 - input_high_end_h : bp["n_fft"] // 2, :
                ]

        X_spec_m = spec_utils.combine_spectrograms(X_spec_s, self.mp)
        aggresive_set = float(self.data["agg"] / 100)
        aggressiveness = {
            "value": aggresive_set,
            "split_bin": self.mp.param["band"][1]["crop_stop"],
        }
        with torch.no_grad():
            pred, X_mag, X_phase = inference(
                X_spec_m, self.device, self.model, aggressiveness, self.data
            )
        # Postprocess
        if self.data["postprocess"]:
            pred_inv = np.clip(X_mag - pred, 0, np.inf)
            pred = spec_utils.mask_silence(pred, pred_inv)
        y_spec_m = pred * X_phase
        v_spec_m = X_spec_m - y_spec_m

        if ins_root is not None:
            if self.data["high_end_process"].startswith("mirroring"):
                input_high_end_ = spec_utils.mirroring(
                    self.data["high_end_process"], y_spec_m, input_high_end, self.mp
                )
                wav_instrument = spec_utils.cmb_spectrogram_to_wave(
                    y_spec_m, self.mp, input_high_end_h, input_high_end_
                )
            else:
                wav_instrument = spec_utils.cmb_spectrogram_to_wave(y_spec_m, self.mp)
            print("%s instruments done" % name)
            wavfile.write(
                os.path.join(
                    ins_root, "instrument_{}_{}.wav".format(name, self.data["agg"])
                ),
                self.mp.param["sr"],
                (np.array(wav_instrument) * 32768).astype("int16"),
            )  #
        if vocal_root is not None:
            if self.data["high_end_process"].startswith("mirroring"):
                input_high_end_ = spec_utils.mirroring(
                    self.data["high_end_process"], v_spec_m, input_high_end, self.mp
                )
                wav_vocals = spec_utils.cmb_spectrogram_to_wave(
                    v_spec_m, self.mp, input_high_end_h, input_high_end_
                )
            else:
                wav_vocals = spec_utils.cmb_spectrogram_to_wave(v_spec_m, self.mp)
            print("%s vocals done" % name)
            wavfile.write(
                os.path.join(
                    vocal_root, "vocal_{}_{}.wav".format(name, self.data["agg"])
                ),
                self.mp.param["sr"],
                (np.array(wav_vocals) * 32768).astype("int16"),
            )


if __name__ == "__main__":
    device = "cuda"
    is_half = True
    model_path = "uvr5_weights/2_HP-UVR.pth"
    pre_fun = _audio_pre_(model_path=model_path, device=device, is_half=True)
    audio_path = "神女劈观.aac"
    save_path = "opt"
    pre_fun._path_audio_(audio_path, save_path, save_path)