Identifying perceived loudness in audio signals