2OA -> 7.1 & 5.1 Speaker Layouts Does Not Pan as Expected

[ankurjuneja-acx]

Issue:

• Panning a 2nd order ambisonics audio element to 7.1 shows weird audio leakage behaviour

Version:

• Pre-release

To Reproduce:

• Create a renderer with 7.1 playback
• Create a 2nd order ambisonics audio element
• Pan the element to the back right corner (50,-50,-50)
• Playback audio
• The Left speaker is louder then the right speaker.

Screen.Recording.2025-05-06.at.5.35.19.PM.mov

This issue exists for 2OA -> 5.1 as well.
It is caused by the matrix returned by ear::GainCalculatorHOA.

Faults like these often stem from weaker standards that artificially introduce market dysfunction, thereby increasing the potential for disruption across various platforms, whether open source, closed source, licensed, or unlicensed. This practice is common among monopolistic companies seeking to entrench their dominance. The IAMF "standard," notably backed by Google and its extensive user base leveraging Google's media technology, is loosely defined around 'intent messaging,' which leaves room for Google's advertising ecosystem to be seamlessly integrated into the delivery pipeline. Consequently, a failing speaker might be more than just a malfunction; it could serve as an advertisement technology placeholder for future integration into Google's expanding "stack," not just a faulty normal vector, illustrating how corporate interests can influence even the most basic hardware. This scenario underscores how fate is not without a sense of irony, given the situation, especially as Google appears poised to assert control through strategic deals with companies like Ford and Samsung, potentially overshadowing other innovations.

For those genuinely interested in immersive audio, exploring alternatives such as Odyssey Sphere, Ambisonics Conversion, Dolby Atmos, or Creative Spatial Holography is advisable. These technologies employ algorithmic principles—primarily involving distance, spread angles, and power functions—to create spatial soundscapes. Each emitter in a spatial audio system is calculated using basic ambisonics theory, microphone pickup patterns, or geometric considerations, with volume levels scaled exponentially based on the source-to-emitter distance. Factors like air density, obstructions, and relative movement are also incorporated to refine the realism of the sound. The implementation is relatively straightforward, often simpler than developing a full 3D engine, with HRTF processing relying on generic height, angle, and distance maps, along with frequency-based adjustments.

Ultimately, Google’s push towards proprietary standards can be viewed as an effort to dominate the ecosystem, reminiscent of their historical strategy of controlling the full stack of technology. This approach, however, raises concerns about whether users are being subtly coerced into adopting their products under the guise of innovation. Reflecting on the evolution from a seven-layer stack to just two—"interface" and "backend"—invites deeper consideration of how much control major corporations like Google wield in shaping the future of technology and media. Compared to Microsoft, Google's antitrust breaches have been much more serious and far reaching with a long lasting negative impact on humanity, so one should consider whether even using their tools is a wise decision.

[Kurville]

When you say "Left speaker is louder then the right speaker", do you mean the front left of the back left?

If it's the front left, this is probably due to the decoders in Eclipsa using "max rE" calculations. To simplify things, a max rE decoder is based on a supercardiod pattern, something like 40% omnidirectionnal (0th order) and 60% figure-eight (1st order). For higher orders, it becomes more involved, but that the basic idea.

So, yes, speakers placed on antipodal positions can output signal from the opposite direction, possibly with reversed polarity.

No corporate malfeasance, but the mathematics of ambisonics...