[AUDITORY] Quantitative roughness data for dichotic pure-tone pairs?

[Sage Betko <sbetko@xxxxxxxxxxxxxx>]

Dear AUDITORY list,

I am looking for psychophysical measurements of perceived roughness for
a steady dichotic pure-tone pair, with one tone presented to each ear
over headphones, as a function of mean frequency, frequency difference,
and presentation level.

For context, I am building a real-time audio visualizer. It currently
applies a Sethares-style pairwise sensory-dissonance heuristic and a
separate Daniel-Weber-style roughness estimator independently to the
two ear signals. A dichotic pair therefore contributes nothing to either
same-ear calculation, although it can produce a binaural beat and, over
part of the beat-rate range, a rough percept.

I have found measurements of binaural-beat frequency and detection
limits (Licklider, Webster & Hedlun, 1950; Perrott & Nelson, 1969),
descriptions of qualitative percept transitions and beat-detection data
around a 500-Hz mean frequency (Ross et al., 2014), and matches
between binaural beats and SAM for modulation salience at 500 Hz
(Grose, Buss & Hall, 2012). Miskiewicz, Rakowski & Rosciszewska
(2006) map perceived roughness for simultaneous pure-tone pairs by
absolute magnitude estimation, but I have not found a reported
dichotic condition. Diotic/dichotic pure-tone work such as McDermott,
Lehr & Oxenham (2010) and Cousineau, McDermott & Peretz (2012)
uses pleasantness, preference, or discrimination rather than roughness
magnitude.

I also found the recent Ichimiya & Ichimiya (2025) preprint, which
includes pairwise roughness judgments for dichotic tonal sequences,
but not a parametric magnitude function for steady tone pairs.

Does anyone know of either:

1. Magnitude-estimation or direct roughness-matching data for dichotic
pure-tone pairs across mean or carrier frequency and frequency
difference, with levels at both ears specified and ideally with
monaural and/or diotic controls; or

2. A validated equivalence between a binaural beat and SAM or same-ear
beating specifically on the roughness dimension?

I considered treating the SAM depth matched to binaural-beat salience
as an equivalent modulation depth in a roughness model. That seems
unjustified: matched modulation salience need not imply matched
roughness, and at the lower rates the salient percept may be motion or
slow fluctuation rather than roughness.

The practical question is whether to leave this cross-ear percept
outside the same-ear calculations or represent it as a separate,
explicitly labeled binaural-interaction cue. I would not apply a fixed
attenuation to a same-ear roughness curve without supporting data.

References and corrections would be very welcome. I will summarize
replies for the list.

Best,
Sage Betko