Frequency-dependent amplitude-panning curves

Today’s most commonly used sound spatialization techniques are based on the 2-channel stereophony. For panning sounds in the stereo image there are two main methods: time-difference and amplitude panning. The latter is mostly preferred by sound engineers due to the better stability of the stereo image. This contribution investigates the frequency dependency of amplitude-panning curves. This dependency had already been tested in the 1960s, where the results stated the amount of lateralization of a sound being affected by its spectrum. However, the listening experiments were performed with a small number of subjects and their results exhibited small consistency across different experiments. With today’s accurate measurement methods, we investigate this frequency dependency again in greater detail through listening experiments in an anechoic room. The direction of a frontal phantom sound source was evaluated by 21 mostly well trained listeners employing a multi-camera infrared tracking system. The phantom source was created by a pair of loudspeakers in a standard stereo setup (+/-30°) that played back the stimuli with different inter-channel level differences. The stimuli were pink noise bursts with various bandwidths: broad-band, low- and high-pass filtered, and third-octave bands at different frequencies. The results prove a significantly larger slope of the amplitude-panning curve for higher frequencies. Finally, the frequency dependency is incorporated in the recently presented generalized tangent law for amplitude panning.