Infrasound is by definition sound too low in frequency to hear. But there are a lot of interesting and competing issues in the questions.
Any speaker can reproduce frequencies of arbitrarily low frequency. What it can’t do is play them loud. As frequency drops, the volume of air you need to move to maintain the same loudness rises, half the frequency, four times the air volume. This is why bass drivers are large and can sweep significant depth. In order to get any useful sound pressure levels they have move air. You phone speaker can easily move air at 20Hz. Just not enough that you would be able to hear it. Whereas at 200Hz, moving the same amount of air it is quite audible.
I would be easy to make a shoebox sized desktop speaker that was flat to 20Hz. The problem is that in order to do so, it would have a maximum loudness that was close to useless.
An obvious example of a very small driver capable of very deep bass is a headphone driver. I have a pair of Etymotic ER-4P. The actual driver in these is not much larger than a grin of rice, yet when properly fitted, these earphones can reproduce prodigious bass, as good as anything you will ever hear. They can do this because they are directly coupled to the eardrum, and the actual volume of air they need to shift is minuscule.
There are other considerations.
Your ear can’t hear below 20Hz. But you may perceive the presence of energy at lower frequencies if they modulate the sound you do hear. Eventually you will hear modulation as a tremolo effect rather than low frequency energy, but there is a cutover.
This leads into a really neat effect. The missing fundamental.
If your ear hears the second and third (and maybe higher) harmonics of a note, it will insert the fundamental. So you could play 40Hz and 60Hz, and you ear would hear the presence of 20Hz. This effect is used to advantage in a range of settings. The very deep notes of a pipe organ actually have the majority of their energy in the harmonics, with maybe 20% or less of their output at the fundamental. The ear still perceives the deep power of the beast. (At low frequencies the brain receives phase information as well as frequency, so is probably able to recognise the harmonics as being in phase and correlated. ) There are effects boxes that create the needed harmonics, and they may be used in recording, performance, and many domestic home theatre systems have the option to synthesise bass this way. It is quite possible to make a phone speaker apparently reproduce deep notes with the same trick.
A noted above, many digital formats are flat to essentially DC. But any compressed audio will almost certainly wipe out and subsonic information. I suspect DSD format won’t hold very low frequencies either. But your bog standard 44.1/16 CD format digital music has the capability, even if the recordings will almost certainly lose it. Tape has an intrinsic low frequency cutoff.