import scipy.io.wavfile


def save_sample(file_path, sampling_rate, audio):
    """Helper function to save sample

    Args:
        file_path (str or pathlib.Path): save file path
        sampling_rate (int): sampling rate of audio (usually 22050)
        audio (torch.FloatTensor): torch array containing audio in [-1, 1]
    """
    audio = (audio.numpy() * 32768).astype("int16")
    scipy.io.wavfile.write(file_path, sampling_rate, audio)