torch-vggish/torch_vggish.py

# Copyright 2021 Zilliz. 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.


import sys
import torch
from typing import NamedTuple
from pathlib import Path
import numpy
import os

from towhee.operator import Operator

import warnings
warnings.filterwarnings("ignore")

class TorchVggish(Operator):
    """
    """

    def __init__(self, framework: str = 'pytorch', weights_path: str=None) -> None:
        super().__init__()
        if framework == 'pytorch':
            import importlib.util
            path = os.path.join(str(Path(__file__).parent), 'pytorch', 'model.py')
            opname = os.path.basename(str(Path(__file__))).split('.')[0]
            spec = importlib.util.spec_from_file_location(opname, path)
            module = importlib.util.module_from_spec(spec)
            spec.loader.exec_module(module)
        self.model = module.Model(weights_path)

    def __call__(self, audio_path: str) -> NamedTuple('Outputs', [('embs', numpy.ndarray)]):
        audio_tensors = self.model.preprocess(audio_path)
        features = self.model(audio_tensors)
        Outputs = NamedTuple('Outputs', [('embs', numpy.ndarray)])
        return Outputs(features.detach().numpy())