Thinking About Apple’s New HomePod mini & Computational Audio
At Apple’s “Hi, Speed.” Special Event on October 13th, which heralded in the new iPhone 12 range with faster, 5G wireless modems, the company also took the opportunity to unveil HomePod mini, a junior version of the HomePod smart speaker.
The original HomePod’s raison d’être was to provide a great listening experience in the home, and this was arguably the device’s strength over similar speakers from companies like Amazon and Google, or even comparable music-oriented offerings from competitors like Sonos. And Apple succeeded. While the HomePod wasn’t perhaps the smartest speaker — Apple’s voice assistant, Siri, was rather limited at the time — it was easily the best sounding. (I reviewed the original HomePod in the June issue of Sound on Sound magazine).
Much has changed in the market for so-called smart speakers since Apple began selling the original HomePod back in 2018, and Apple themselves have made huge advancements in the audio technology used in the company’s products in the intervening time. Therefore, this new member of the HomePod family undoubtedly benefits from these developments in both its hardware and software, which we’ll discuss in this initial examination.
Magnetic Attraction
The heart of the HomePod mini’s speaker is based on an Apple-designed full-range driver that uses an “incredibly powerful” neodymium magnet. Such magnets are among the most powerful permanent¹ types available, and they’re particularly useful in smaller speakers because they pack a large amount of magnetic energy in a tiny package. This enables a reduced-size driver to deliver significant performance with crisp high frequencies and low distortion across the spectrum. Apple claimed the original HomePod driver could deliver 20mm of excursion², although currently no similar information has been published for the mini’s driver.
Complementing the driver is a set of force-cancelling passive radiators that are used to extend the bass response in a tight and controlled manner, which can be observed on either side of the HomePod mini’s driver. The passive radiators provide bass frequencies by using air pressure generated by the driver, which is why they’re described as being passive as opposed to active (where an independent power source would be required). This provides a decent amount of bass using relatively little power, and effectively simulates a larger subwoofer given that the air pressure created can only be released through the radiators.
In order to prevent the HomePod mini from rattling around when reproducing low frequencies, force cancelation is utilised by the radiators so that the force generated by one of them is cancelled out by force generated by the other. This helps to maintain a tightness in the bass spectrum by minimising the vibrations that would otherwise occur and interfere with the system. Apple has used force-cancelling technology before, most notably on the 2019 16-inch MacBook Pro, and this resulted in a particularly impressive bass response for a laptop audio system without exciting the surface on which it was placed.
Rounding out the HomePod mini’s speaker hardware is an acoustic waveguide, which creates a 360-degree sound field that, like the original HomePod, provides a consistent listening experience no matter where you or the mini are located within a room.
As the name implies, an acoustic waveguide is a physical object that guides audio waves. A simple example of a waveguide would be a pipe, where the audio waves are ushered from one end of the pipe to the other rather than simply radiating out of the driver, which would make it hard to focus the sound and lead to a loss in acoustic energy from the listener’s perspective. Apple’s custom-designed waveguide pushes the sound out from the bottom of the HomePod mini speaker in order to create the consistent 360-degree sound field that’s required.
¹ A permanent magnet is an object made from a material with a hard structure that maintains a persistent magnetic field.
² Excursion is the linear vertical distance a driver can move from its initial, resting position, dictating the amount of air that can be pushed by the driver and therefore the maximum amplitude.
Software Eats the Speaker
The hardware design of the HomePod mini’s speaker is certainly neat, especially for such a small device. However, the real magic orchestrating the behaviour of the various speaker components is the signal-processing software Apple’s engineers have developed to fully optimise the overall audio performance.
Where the original HomePod was powered by Apple’s A8 system-on-a-chip that was developed for the iPhone 6, the HomePod mini is based around Apple’s S5 system-in-package instead, which was first seen in the Apple Watch Series 5. Amongst other duties, this handles the HomePod mini’s audio processing, and Apple refer to this type of processing as computational audio to describe how software can enhance the reproduction hardware ‘s functionality. A corollary to this is what Apple term as “computational photography” in the iPhone, where software is used to enhance the results that can be captured by the device’s camera system.
Using a combination of both the inputs from the built-in microphones and the sound source being played back, Apple say that the HomePod mini can apply real-time processing — that is, computational audio — to provide balanced, finely-tuned sound at any volume. As explained during the keynote, this works by analysing the audio signal to be played back, and then tuning the loudness and dynamic range accordingly to get the best result from the driver and radiator array when the resulting signal is reproduced. This processing occurs 180 times a second, which means it operates with approximately 6ms of latency. And since this is a smart speaker, noise cancellation is employed so that decipherable voice commands can be issued to the HomePod mini via “Hey Siri” — even when the speaker is playing back audio or used in a bustling environment.
The concept of computational audio, while not new in the world of speakers, is a clever way of using software to dynamically get the most from smaller and cheaper components in any situation. Speakers can be incredibly expensive, and high-end quality is usually achieved by elegant hardware designed to respond a certain way in the analogue domain given an ideal environment. Therefore, to repeat the comparison made earlier, in much the same way computational photography gives you results that rival cameras with larger and more expensive components, so to does computational audio in its domain.
Another example of this approach is that, as became possible via a software update to HomePod after its initial release, you’ll be able to use two HomePod minis in a single configuration. In this setup, the output of both speakers will be detected and automatically balanced to provide a stereo-like sound field. The advantages of software!
Wait & Hear
Ultimately, the proof of the HomePod mini pudding will be in the listening rather than the eating, and it will certainly be interesting to hear how the technology described here translates into an audible experience. So far, the only downsides seem to be those inherited from the original HomePod: firstly, that the power cable isn’t detachable (despite now being USB-C based); and secondly, as is both inevitable and consistent with Apple’s wireless vision of the future, there’s no hardware audio input, which I still think is a shame.
- HomePod mini will be available for pre-order on Friday, 6th November for just $99, with availability beginning 10 days later.