Improved Localization for Binaural Recordings and Stereo Program Material Using ‘Blumlein Shuffling’
University of Derby
University of Derby
University of Derby
2025-11-07
Improved Localization for Binaural Recordings and Stereo Program Material Using ‘Blumlein Shuffling’
Author / Speaker: Jonathan J Digby, University of Derby
Co-authors: Dr Adam J Hill & Dr Bruce J Wiggins, University of Derby
An example spherical-response microphone polar pattern
Schoeps MK2 omnidirectional microphone manufacturer predicted polar response.
\(\rule{10cm}{0.5pt}\)
For an angled pair of near-coincident omnidirectional microphones (\(\pm 45...55^{\circ}\)), a typical spherical response can provide Inter-microphone Level Differences which emulate Interaural Level Differences (ILD).
This can be effective for source angles of incidence up to around \(\pm 20^{\circ} \dots 30^{\circ}\).
Beyond this a greater effect is needed.
A hybrid baffled microphone technique
Variable baffle-size calculator for near-coincident microphones (R script)
Referenced ILD targets : Baffled near-coincident microphone pair
The interaural level difference plotted as a function of source azimuth (Incident angle) for tones of six different frequencies, as computed in the spherical-head model with antipodal ears. All six curves have a peak, but it is hard to see for 500 Hz.
Macaulay, Hartmann, and Rakerd, Mar. 2010 [6,fig 1]
2x Schoeps MK2, 200 mm, +/- 45 degrees;
Baffle: 222 mm diameter, 90 mm thick
Reproducing the ‘binaural’ signal using loudspeakers at an oblique angle
For the evaluation of stereo impression, a listening position on the perpendicular line passing through the middle of the stereo base is preferable.
Admissible limits of base width, b:
2,0 \(<\) b \(<\) 4,0 m
h \(\approx\) 0,9 b
\(\theta\) = 60\(^\circ\)
r\(_L \le\) 0,8 m
Reference listening position
about 1,2 m above the floor
Blumlein’s shuffling matrix
Assuming the original currents differ in phase only, the current in the difference channel will be \(\frac{\pi}{2}\) different in phase from the current in the summation channel. This difference current is passed through two resistances \(d\) and \(e\) in series between which a condenser\(^{\dagger}\) forms a shunt arm. The voltage across this condenser \(f\) will be in phase with that in the summation channel. [11,p 11]
Signal Analysis \([\textit{MATLAB}]\)
Left channel signal is fed to the Right input with a delay and level adjustment.
This approximates the ILD and ITD offsets for a distant oblique source at azimuth \(\theta = 50^{\circ}\)
Sine bursts, raised-cosine envelope. Timescales (x-axis) adjusted to match signal frequency.
Filtered Difference: 500 Hz, low-shelf, +20 dB; Filtered Sum: offset frequency, low-shelf, −1.5 dB
Example recording setup
Microphone array, baffled:
Schoeps MK2 microphones,
spherical response, matched pair
190 mm spacing, \(\pm\) 45 degrees
1,8 m above the floor
median circular baffle:
acoustic foam[4],
190 mm diameter, 90 mm thick
Microphone array, unbaffled:
Sennheiser 8030 microphones, figure-8 response, matched pair
200 mm spacing, \(\pm\) 0 degrees
1,7 m above the floor
Inverse shuffling matrix emulations
Left channel signal is fed to the Right input with zero delay and -20 dB level adjustment.
This emulates an amplitude panned source.
Sine bursts, raised-cosine envelope. Timescales (x-axis) adjusted to match signal frequency.
The 20 dB offset results in a ‘shuffled’ delay at 250 Hz of \(\approx 0,33\) ms.
Electro-Acoustics Research Lab
