Perceptual Effects of Dynamic Range Compression in Popular Music Recordings - January 2014
Accurate Calculation of Radiation and Diffraction from Loudspeaker Enclosures at Low Frequency - June 2013
New Measurement Techniques for Portable Listening Devices: Technical Report - October 2013
Minimally Audible Noise Shaping
Normal quantization or requantization noise is white, but the ear's sensitivity to low-level broadband noise is not uniform with frequency. By adopting a suitable weighting curve to represent low-level noise audibility, one can design dithered requantizing noise shapers to approximate the inverse of the audibility curve and hence achieve the least audible noise penalty. If Fielder's modified E-weighting curve is adopted as a model of the 15-fon audibility curve, a reduction of 10.9 dB in perceived noise is possible by the use of a simple second-order noise shaper. This result is already within 0.6 dB of the theoretical minimum set by information theory, and almost a 2-bit gain in apparent signal-to-noise ratio. Even greater perceived noise reductions are possible if one adopts an audibility weighting curve which more closely approximates the ear's precipitous high-frequency rolloff, and incorporates a higher-order filter into the noise-shaper's feedback loop. In face, a 20-dB apparent reduction in the requantizing noise is then possible with filters of modest order, but the penalty is a significant increase in the total noise power. This paper explores these questions and illustrates some of the available options. Such noise shaping will soon get advantageous in order to preserve on the 16-bit compact disc the lower noise floor of an original 18- or 20-bit master recording.
Click to purchase paper or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $20 for non-members, $5 for AES members and is free for E-Library subscribers.