HomePod Technical X-ray

When Apple announced the HomePod, I knew instantly that I wanted it. I hadn’t heard it, I didn’t know if I was going to like it and i was a Spotify user, but an Apple speaker focused on music rather than the voice assistant was a big temptation for me. In fact, I needed such a product (a music speaker) because listening to podcasts and music with the iPhone on top of the kitchen counter did not convince me lately and I was considering buying at some point a Sonos or some similar speaker.

Story of an audiophile

My predilection for sound goes beyond a pure hobby as I have dedicated myself professionally to it for almost 30 years. It was in 1991 when on my first PC, unolivetti PCS 286, I committed «the madness» (for that time) of spending 39,900 pesetas (which today would be just under 240) to buy a Sound Blaster Pro sound card with digital synthesizer that produced sound similar to a sint analoger, compatible with Adlib cards (one of the first MIDI sound standards for PC, which used a Yamaha YM3812 sound chip and created music sounds by FM synthesis) and also incorporated 2 8-bit stereo digital channels at a time 22,500Hz (if played in a monoaural it reached 44,100Hz.). To give you an idea, a CD is registered at 16 bits and 44,100Hz.

In case you doubt what those figures we hear so much about bits, hertz and so on mean, clarify that what today is called CD quality uses integer values at 0 and 65535 (2 raised to 16) to digitally represent a point of a sound wave. In a second, points are recorded 44,100 times (the hertz of their frequency) and with that we have the standard of digital music. Imagine that the first sound card used only 8 bits, so instead of recording the wave with values between 0 and 65536 it between 0 and 256 (2 raised to 8). A quality not very good, but that at the time was magic for me.

Since then the sound has evolved a lot and in my career I have come to test and work with sound devices of such emblematic brands as Gravis or Turtle Beach and I have been lucky enough to see how digital sound has been improving year by year over the last 30 years and has evolved to qualities that allow almost any Apple computer of the last 20 years can serve for professional sound production with programs of extraordinary quality like Protools, Logic Pro X or Nuendo.

-Digital sound and music production have evolved a lot in the last 30 years and today (and for a few years now) we can have a house sound studio with professional production quality and result-

In fact, it was music that took me to Apple, when many years ago I visited film music composer Sergio Moure at his home, and was composing for a TV series no less than with a PowerMac G5 and the help of Logic Pro 7. It was supposed to be 2005, if I remember correctly. He had the series in a window on an auxiliary monitor and then Logic full of digital sound tracks recorded from him playing guitar or other collaborators playing real instruments, attached to synth tracks real instrument software. I remember there were over 40 tracks on the screen and everything was running smooth and in real time. At that moment I fell in love with that computer and knew that if I wanted to take the next step in sound and music I should buy an Apple. I had always known that music and Apple have gone hand in hand and that media professionals have always used these machines for their simplicity and that «it just works». But watching it live was a very important thing. If we look at the evolution of music and sound production, the beginnings were more united to the famous Atari ST, but when it was disappeared it was Apple who took the witness to the music production software.

Apple, music and sound have always gone hand in hand and for that reason, I shouldn’t miss the opportunity of a speaker made by Apple with the highest musical standards, even though I had my reluctance that their sound wasn’t balanced enough and went It’s dangerously towards the dark side of Beats, with excess bass, lack of treble and unbalanced sound. I respect whoever likes it, but in the balance of frequencies is the best sound quality. And the challenge for Apple was no small.

-For the professional sound community, the purchase of Beats by Apple was not a good sign because from a technical point of view, beats headphones are far from what we can call «musical quality». Obviously, everything is relative, but the sound professionals agreed on that analysis. Luckily, Apple was left «alone» with the important thing about Beats: its streaming service.-

To do this, as only Apple knows how to do, he borrowed different ideas spread across different technologies or devices and made them his own to create the best musical speaker to date (at least in the opinion of the writer).

Testing the HomePod

I’ve had the HomePod for a few months because of the fortune my parents went on a trip to New York and I asked them for the favor if they could bring it to me. As soon as I turned it on, my first surprise was the extraordinary simplicity of the setup. Plug in, boot ed in a few seconds, Zoom in on the phone and automatically and with a few taps were already transferring all the settings of WiFi, Apple ID account and so on to the device. Knowing that I would have it, a day before I signed up for Apple Music (I was a Spotify user) and with the help of the magnificent Songshift app, I was able to migrate all my playlists from one service to another with a hit close to 95% on almost 2,000 themes that make up the set spotify (about 8 years as a user of the service).

-If you want to migrate songs between lists of different streaming services, use Songshift. There is no app that works better

Set up your HomePod, it suggests a series of phrases to start using and in the last step you are invited to say «Siri, play some music». I said it and Siri told me: «Okay, Julio. Let’s start with something from Michael Giacchino.» Good taste, I thought. I had already kept an eye on my lists and subscribed artists and had an idea of what I was hearing. And the music began. A theme from the «Star Trek» soundtrack. The first impression was very good. Good spatiality, better than a common speaker. The sound doesn’t give you the feeling that’s coming out of the speaker. It’s obvious that you know he’s doing it, but the sound feeling is that the sound is around the device but its source is not the same.

The bass, well marked but not exaggerated. The treble with good presence, the means clear. The frequency distribution was excellent. It’s not the sound quality of my Onkyo equipment, obviously, because we’re talking about a 349-degree speaker and it can’t be compared in circuit quality, connections and speaker size to a THX 7.1 certified Onkyo like the one I have at home. But it’s certainly the best you can hear (and I’ve heard) in that price range. But hey, it was just a song. I called my daughter and she asked me to put on Vaiana’s soundtrack. After recalling that in English the album is Moana (the name had to be changed in Spain due to rights problems with the name Moana that is registered in the country) we asked for it and it began to be heard. The Hawaiian percussions of the first track and the choirs had extraordinary strength. We turn up the volume, a lot, to test its capacity… not a simple distortion of the output. A clean sound. The excellent music of Lin-Manuel Miranda in all its glory.

I was going to ask Siri for another song and my first thought was to turn down the volume because otherwise Siri wouldn’t hear me safe with the cacophony that would be created between the music and my voice. But I preferred to try that magic that they said by which Siri heard you at any distance. I parted and said in a tone that I didn’t even hear myself, «Hey Siri, play some music from John Williams.» He heard me without any problems. How is it possible, will some say? Because I can guarantee you it looks like magic.

Frequency issue

It’s all about frequencies and the use of microphones. HomePod has an array (a collection) of microphones that surround it, where each of them specializes in a task. They are divided into two groups: the microphones that hear the music that HomePod itself plays and those that are created to capture the frequencies of the human voice.

-The microphones included in the HomePod have different functions. Some listen to what HomePod itself plays and others focus on capturing only the frequencies of the human voice.

The wave spectrum of a sound that humans can capture (on average) varies between 20Hz and 20,000Hz. The 20Hz represents the lowest (or bass) pitch we can hear and the 20,000Hz the sharpest. Every human being has more or less capacity, obviously, but that range is what represents what most devices usually emit. Between 250 and 6,000, more or less, is where most of the sounds we pick up through the ear are placed. A microphone, depending on its configuration, is configured to pick up sounds over a given frequency range. But from all that spectrum, the human voice occupies very specific ranges. Its basic frequency or lower range is typically between 85 and 180Hz for a man’s voice and 165 to 255Hz that of a woman. From there, each voice can get more or less up with limits that typically reach 3,000Hz on average. That’s what’s often considered the limit, usually. That means that if I have microphones adjusted to hear those frequencies better and not higher or lower ones, they will better pick up the voice and automatically discard (they won’t hear) what is not human voice. So when Apple microphones hear our voice, they look for it right on its frequency.

But also, to help in understanding, the rest of the microphones also listen. What do you hear? The sound that the speaker itself generates. If we have a voice in a song or we’re listening to a podcast, those voices will go on the same frequency. But HomePod knows that these voices are being reproduced by him and through sound-treatment algorithms he removes them from what he hears and isolates only the voice that does not come from himself: ours.

Music is Machine Learning

To understand this better and what function the HomePod fulfills we need to talk about one of the most criticized features of this speaker: it has no equalization. We cannot alter the response of the frequencies to change the different frequency bands of the sound by means of filters and thus enhance bass, midors or treble. The HomePod is equalized and adjusted only depending on where it is placed and how the sound plays and is returned to its own microphones. It is a process governed by Machine Learning algorithms that have a trained model capable of recognizing, depending on how the speaker listens to itself, how it should adjust this equalization automatically on each of its speakers, in order to create the best special and balanced effect of the music.

-The HomePod listens to itself and using Machine Learning models, modifies how the music is heard to suit the sound needs of each style or song, as well as the acoustic needs of where it is being heard: computational sound-

This system is based on the one that some Home Cinemas have to adjust themselves the first time we configure it through an external microphone calibration. A system by which a special microphone is put in the center of the room and speaker settings and the system itself plays sounds at different volumes and frequencies to detect the distance to which each of the speakers is, measure and adjust its levels automatically and equalize in the best way. It is very important to understand this: not even a home cinema of thousands of euros (what we would call an A/V Receiver) has an equalizer. In fact, an equalizer is something that is not recommended in the hands of non-expert users because it can spoil the same sound too much. However, these equipments pick up the frequencies of a sound source and adjust themselves for the best and most balanced sound. They prioritize the sound mix that the source has to the settings one wants to put.

In fact, in the case of HomePod, the EQ profile settings for iTunes or Apple Music songs are recognized and applied. It doesn’t sound the same on a HomePod on Katy Perry’s «Swish, Swish» with Nicky Minaj (my kids ask for it, I can’t do anything about it), that that music masterpiece called A-Ha’s «Take on Me» or even Queen’s indescribable Bohemian Rhapsody in each of its different musical parts that emphasize different styles.

-A HomePod is not a musical speaker to use: it is a device that applies computational sound to improve the quality of it through Machine Learning.A step further in musical devices-

This is essential to understand the HomePod: the Apple speaker is not a speaker to use, it applies the concept of computational sound. Just like the quality of mobile cameras depends today more on the software that processes the information that comes from the camera sensor, which is why an iPhone XR or a Pixel 3 does wonders in their photographs (not because they have better cameras , but by the software that processes what this captures) the HomePod plays the sound computerfully and does not play the same theme like John Williams’s «Across the Stars» for Episode II, the Imperial March, a Hans Zimmer theme for Gladiator or the soundtracks of John Williams Inception or Interstellar. Inception’s «Mombasa», for example, is a great example of how far the HomePod can go in sound clarity despite the musical violence of the theme. In fact, let’s focus on another hans Zimmer work to exemplify what homePod does with sound: the soundtrack to Interstellar, one of my favorites.

Lossy, lossless, CD and 24-bit (HD)

I have 3 editions of «Interstellar»: one of them is the Version of Apple Music, another on CD and a last one on 24-bit sound. The first has a quality of 256kbps. This means that Apple uses a constant bandwidth of 256 kilobits per second for the data that makes up the sound. Every second of music takes up a data stream just that bandwidth. To get an idea of the amount of information, let’s think that 1,024 kilobits would be a megabit. And we already know that we already have 300 or 600 fiber connections, right?

The music on Apple Music is in lossy quality which means that to be compressed so that it can be transmitted in that bandwidth, it has removed sound information that «is considered» that the ear does not capture but is stored in the original track. This subtracts what audiophile experts call «matics.» Yes, I know that we enter dangerous terrain where there has been talked about a well-compressed song in MP3 or MPEG4 (the algorithms that Apple uses in its Music service and in iTunes) is indistinguishable by any ear of a CD-quality track without that «loss d and information.»

But don’t be fooled. If I intend to compare sound tracks with or without compression in your data, I can’t do it either way. If you use headphones (headphones) whose ability to play sound frequencies does not allow you to play precisely those frequencies that have been eliminated when converting the music in this lossy format, we are sure that you would hear indistinguishable a compressed sound from one that is not. To capture these nuances we must use speakers (speakers) or headphones (headphones) that have a good quality that allow to play those frequencies that have been erased to compress the music. That is, if we put music on CD and on lossy as an MP3 or MP4, we have to hear it on a good computer that is able to play the frequencies that make it distinguish between the compressed sound or not. Otherwise, it’s impossible for us to tell each other apart.

-If you use headphones (headphones) whose ability to play sound frequencies does not allow you to play precisely those frequencies that have been removed by compressing the music, you would indistinguishably a sound compressed from one that is not-

Hearing this difference means using professional brands and midranges (at a minimum). Headphones of about 200 euros onwards (wired, always wired) or professional speakers with a good finish. It’s no use. Not even some AirPods. In fact, wireless music is already compressed to be broadcast. All codecs (or encoders-decoders) of sound via Bluetooth are lossy compression. The closest thing to one that doesn’t lose quality is the aptX standard via Bluetooth (which doesn’t support Apple). But if we use a speaker like the HomePod that uses WiFi, there we do have bandwidth to transmit more quality.

The HomePod allows you to distinguish the difference in quality of a sound, and that says a lot in its favor in terms of quality. And a good proof of this is to compare the highest quality copies of the Interstellar soundtrack: CD and 24-bit. In this case 24 bits and 48Khz. 24-bit sound is a very significant improvement in music. It means that to represent a sound, not only samples are collected 48,000 times per second, it is that the value is a number in 0 and 16,777,216 (2 raised to 24). We went from having just under 70,000 values to represent how sharp or serious a sound is, to having almost 17 million possible values. To say that the difference is indistinguishable from that empirical data, does not make much sense as we can understand. Another separate discussion is whether a human ear is capable of grasping that difference, which is a unresolved discussion.

This quality, which is used in studios when it comes to producing music and then reduced for commercial distribution to 16 bits, allows to offer much wider sound nuances and incredible quality. As long as we have a speaker or system that allows us to hear it, like a HomePod. Again, headphones of less than $200 or a speaker like Amazon or Google’s (except Google Max) won’t work. In fact, the sound standards of Blu-ray movies or UHD discs (Dolby True HD, DTS-HD or even the latest Dolby Atmos or DTS-X) are in 24 bits of sound in lossless or lossless compression. A lossless format is one that compresses the data but does not alter the original sound wave so it does not lose quality. It’s like «making a ZIP» to the music and taking it out of the ZIP in real time to play it (for a more approximate idea of this concept). In fact, HomePod supports the FLAC format of music in lossless compression (lossless) as well as Apple’s own format for this type of music: the ALAC (Apple Lossless Advanced Codec).

How do I listen to 24-bit music on a computer that allows playback? The comparison is simple: Apple Music sounds like radio. You can hear it. That’s all right. A CD is listened to as a recording of the music, with more nuance and quality. A 24-bit master is like being present at a concert where they play that live music. I’m not talking about an amplified concerto. I’m talking about you guys listening to an orchestra. That’s the feeling. And some of you will be wondering, but does the HomePod play 24 bits of sound? Yes. Up to 24 bits and 48Khz.

But be it: this is the sound quality that one can launch through AirPlay 2 from a FLAC source with Quicktime or VOX Player from a Mac. However, it should not be forgotten that Apple clearly states that sound can be compressed in its transmission through the protocol. But I do have to say that you can tell the difference: you can see between the soundtrack on Apple Music and the CD (it’s very clear, for example, in the wind blows on some tracks or in the clarity of the organ sound, where it itches the output in some stretches (you hear saturated volume n) that then not on the CD. And when you put it in FLAC (which for now is only supported by those two Mac apps I eat) the clarity and sharpness of the sound is even greater and no matter how much you turn up the volume on the more sound parts, it doesn’t lose sharpness within its strength. And all the high-pitched nuances and sound wall effects that Hans Zimmer uses to make instruments more loud, they sound completely clean. Although they can be heard at high volume, their sharpness is absolute.

-Compressed music with loss espo is heard as the radio. A CD is listened to as a recording of the music, with more nuance and quality. The 24 bits is like being present at a concert where they play that live music

Using Machine Learning and listening to how different frequencies coming out of the speaker are returned, the HomePod is equalized and adjusts in real time to the place where it is. Do not forget that being plugged in, you do not have to limit your process capacity and with it you get a better sound because depending on how you listen to yourself you are able to guess the acoustics of the room or place where you are and adjust your sound to it for a greater entfacility and quality.

Before I finish, I want to clarify something: Some readers have jumped my neck by stating that a sound in compressed format such as MP3 and a CD differ and that the compression of data is done to the sound, not only compressing the data but eliminating frequencies that «it’s supposed» that the human ear doesn’t pick up, can be differentiated. I give an example: to me a person can teach me different shades of white: broken, egg, raw… and that person will distinguish them perfectly. I don’t have the ability. My vision won’t allow me to. In the same way each has a hearing ability, some better than the other. And in the same way, a speaker of 50 o, as one of 200, another of 600 or one of 6,000 . I say and affirm that I note the differences with the HomePod, but it is my opinion and my perception of an empirical test. Another person in the same test (my wife, without going any further) will not be able to tell it apart. That doesn’t make any less true my experience or yours.