3-D Sound and Spatial Audio

MUS_TECH 348

What are Some Options for Creating DTFs?

Pole-Zero Design of Filters

4th-order filters designed to capture the essential features of HRTFs

Kendall & Rodgers 1982

Pole-Zero Design of Filters

DTFs are created with low computational overhead.

Kendall & Rodgers 1982

Pole-Zero Design of Filters

Essential features are encoded in look-up tables.

Kendall & Rodgers 1982

Pole-Zero Design of a Ensemble of DTFs

8th-order filters interactively designed so that poles and zero migrate smoothly between the specified points.

Kendall 1985

Pole-Zero Design of a Ensemble of DTFs

Filter response is different but the relative pole and zeros positions have shifted in small amounts.

Kendall 1985

Principal Components Analysis and

Resynthesis Provides an Analytic Approach

to Simplified Filter Design

Martens 1987

PCA Analysis produces a reduced set of ‘basis filters’ that capture all of the significant variation in the data.

Principal Components

Novel DTFs can be synthesized from a combination of basis filters, each with a different weight.

Martens 1987

What are the alternatives to Schroeder Crosstalk Cancellation?

“Head Tracked 3D Audio Using Loudspeakers”Bill Gardner

• Head tracking with loudspeakers

• Do crosstalk cancellation below 6kHz

• Designed for computer systems

“Head Tracked 3D Audio Using Loudspeakers”

Q Sound - Pro AudioDanny Lowe and John Lees“Sound Imaging Process” US Patent 5,046,097

Reduction to Practice

Empirical Approach to Design

Q Sound

Results of measurements are consolidated in look-up tables.

How does this work?Crosstalk happens!

Sum of two sine waves of the same frequency?

It’s a sine wave:

sin(2 π f t + Ø1) + sin(2 π f t + Ø2) =

2 cos[0.5 (Ø1 - Ø2)] sin[2 π f t + 0.5 (Ø1 + Ø2)]

Constant that changesamplitude Sine wave with different phase

Q-Sound

What arrives at the ears is the result of the addition.

No crosstalk cancellation!

Q-Sound

Advantages/Disadvantages

• Images not so sensitive to seating position – broader sweet spot

• Listeners can turn their heads toward the image

• Works best with loudspeakers close together

• Difficulty with elevation.• Timbre coloration.

Similar Approaches Taken By

Cooper-Bauck

Auris (Kendall)

Using other model-based approaches

Loudspeaker DTFs Based on Perceptual Model

What loudspeaker signals were required to produce ‘central spectrum’ and ‘interaural disparity’ for a given direction?

Cross talk was incorporated into the computation. The DTFs did not resemble HRTFs.

Kendall 1988

Loudspeaker DTFs Based on Perceptual Model

Path of sound motion degrades gradually with off-center seating.

General Observations(Given a “good” reproduction environment)

What simulated sound locations work most reliably?

Horizontal plane

General Observations

Frontal plane

Distance

Central rear

What simulated sound locations work most reliably?

Sound Examples

http://musictechnology.northwestern.edu/3dsound/gallerytop.html