use test offset in audioarray.compare for all steps

        Set to True to only apply the threshold comparison for the reference signal

        for whether to include a segment in the ssnr computation. Use this to align

        behaviour with the MPEG-D conformance specification.

    test_start_offset_ms: int

    test_start_offset_ms: (non-negative) int

        offset in miliseconds for test signal. If > 0, the corresponding number of samples

        will be removed from the test array like so: test = test[sample_offset:, :], default 0

        will be removed from the test array like so: test = test[sample_offset:, :].

    Returns

    -------

    result: dict

        Comparison results

    """

    framesize = fs // 50

    if test_start_offset_ms < 0:

        raise ValueError(

            f"Test_start_offset_ms has to be non-negative, but {test_start_offset_ms} was given."

        )

    test_start_offset_samples = int(fs * test_start_offset_ms / 1000)

    diff = abs(test[test_start_offset_samples:, :] - ref)

    test = test[test_start_offset_samples:, :]

    framesize = fs // 50

    diff = abs(test - ref)

    max_diff = int(diff.max())

    result = {

        "bitexact": True,