Listening test data collected in 2003 on 49 audio program samples were used to formulate the ITU-R BS.1770 program loudness prediction algorithm. The validity of these data at low frequencies was unproven. Octave-band analysis has therefore been performed on the test samples to check for audibility in each band. Results suggest that further listening tests may be needed to obtain reliable low-frequency data. A multiple regression analysis was also performed on the octave-band data to obtain a least-squares weighting curve for comparison with the BS.1770 / RLB2 weighting curve. Results suggest that while the BS.1770 curve performs well, there is still room for improvement.
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.