mirth

def squeeze_collate(batch):
    batch_x, batch_y = list(zip(*batch))
    batch_y = np.array(batch_y, dtype=np.float32)
    batch_y = torch.from_numpy(batch_y)
    batch_x = torch.cat(batch_x, dim=0)

    return batch_x, batch_y

mirth

class NoiseDataSet(Dataset):
    def __init__(self, X, y):
        super().__init__()
        assert(len(X) == len(y))

        self.X = X
        self.y = y

    def __len__(self):
        return len(self.X)

mirth

import pandas as pd
from pathlib import Path

def get_ytid(x):
    name = Path(x).name
    ytid = name[: name.rfind('_')]
    return ytid

maybe_noise = pd.read_csv('Engine+noises_2020-08-24T19_34_31.csv')
maybe_healthy = pd.read_csv('Engine+healthy_2_2020-08-24T22_17_40.csv')

mirth

class NoiseClassifier(nn.Module):
    def __init__(self):
        super().__init__()
        self.vgg = vggish(preprocess=False, postprocess=False)
        self.vgg.embeddings = nn.Sequential(
            nn.Linear(512 * 4 * 6 * WINDOW_MULTIPLIER, 256),
            nn.ReLU(True),
            nn.Linear(256, 256),
            nn.ReLU(True),
            nn.Linear(256, 256),

mirth

from itertools import imap, ifilter

class map:
    def __init__(self, callable):
        self.callable = callable

    def __ror__(self, *iterable):
        return imap(self.callable, *iterable)

class filter: