import numpy as np

def norm(X_pad):
    mean_x = X_pad.mean()
    var_x = X_pad.var()
    return np.array([(x - mean_x) / np.sqrt(var_x + 1e-7) for x in X_pad])      
      
      
def pad(x, max_len=64600):
    x_len = x.shape[0]
    if x_len >= max_len:
        return x[:max_len]
    # need to pad
    num_repeats = int(max_len / x_len) + 1
    padded_x = np.tile(x, (1, num_repeats))[:, :max_len][0]
    return padded_x


def pad_random(x: np.ndarray, max_len: int = 64600):
    x_len = x.shape[0]
    # if duration is already long enough
    if x_len > max_len:
        stt = np.random.randint(x_len - max_len)
        return x[stt:stt + max_len]

    # if too short
    num_repeats = int(max_len / x_len) + 1
    padded_x = np.tile(x, (num_repeats))[:max_len]
    return padded_x