Generalized Metrics for Constant Directivity

Sridhar R, Tylka JG, Choueiri EY. Generalized Metrics for Constant Directivity. J Audio Eng Soc. 2019;67(9):666-678. doi:10.17743/jaes.2019.0022

@article{Sridhar2019_20543,
  author = {Sridhar, Rahulram and Tylka, Joseph G. and Choueiri, Edgar Y.},
  title = {{Generalized Metrics for Constant Directivity}},
  journal = {Journal of the Audio Engineering Society},
  volume = {67},
  number = {9},
  pages = {666--678},
  year = {2019},
  month = sep,
  publisher = {Audio Engineering Society},
  doi = {10.17743/jaes.2019.0022},
  url = {https://doi.org/10.17743/jaes.2019.0022}
}

TY  - JOUR
TI  - Generalized Metrics for Constant Directivity
AU  - Sridhar, Rahulram
AU  - Tylka, Joseph G.
AU  - Choueiri, Edgar Y.
T2  - Journal of the Audio Engineering Society
J2  - J. Audio Eng. Soc.
VL  - 67
IS  - 9
SP  - 666
EP  - 678
PY  - 2019
DA  - 2019/09/06
DO  - 10.17743/jaes.2019.0022
UR  - https://doi.org/10.17743/jaes.2019.0022
PB  - Audio Engineering Society
LA  - en
AB  - Many applications in audio benefit from transducer arrays whose directional characteristics do not vary with frequency, as for example sound reinforcement and selective microphone beams. The coverage angle should be constant over a usable frequency range. Metrics are proposed for quantifying the extent to which a transducer’s polar radiation (or sensitivity) pattern is invariant with frequency. As there is currently no established measure of this quality (often called “controlled” or “constant directivity”), this paper proposes five metrics, each based on commonly-used criteria for constant directivity: 1) a Fourier analysis of sensitivity contour lines (i.e., lines of constant sensitivity over frequency and angle), 2) the average of spectral distortions within a specified angular listening window, 3) the solid angle of the frontal region with distortions below a specified threshold, 4) the standard deviation of the directivity index, and 5) cross-correlations of polar responses. These metrics are computed for ten loudspeakers, which are ranked from most constant-directive to least, according to each metric. The resulting values and rankings are compared, and the suitability of each metric for comparing transducers in different applications is assessed. For critical listening applications in reflective or dynamic listening environments, metric 1 appears most suitable, while for such applications in acoustically-treated and static environments, metrics 2 and 3 may be preferable. Furthermore, for high-amplitude applications (e.g., live sound) in reflective or noisy environments, metrics 4 and 5 appear most suitable.
ER  -