We can't have sidetone nor full duplex in wide area cellular telephony because end-to-end latencies become so long the far-end echos go beyond asthetically disconcerting.
Try this just for fun: Call someone right next to you via your cell phones and talk to each other. One finds one can speak several words, maybe even a short but complete sentence before the other person starts hearing any of it. It can become a laughably long latency when experienced so directly.
But this cumulative latency ( be it from queuing delays, voice packet grouping for burst transmission, buffering delays along the many packet switches from here to there, you name it ) comprise engineering trade-offs necessary for optimal radio-level multiple access, efficient packetizing, and economically efficient (packet) transmission.
In other words, I'd postulate we can't (today) afford full duplex, sidetone-included, wideband (5KHz+) sub-20-millisecond latency in cellular telephony. Cellular is a different thing, but surely it's its own kind of magic.
But consider modern office telephony over Ethernet: I don't know the internals of it, but aesthetically, I find it significantly better than the analog telephony I remember from 40 years ago: The sidetone's there, it's full-duplex, speakerphone functionality is absolutely superb, and subjectively I find it sounds good.
I'd guesstimate a three orders of magnitude latency ratio between LAN-based telephony vs cellular telephony seems to me key to the very different engineering possibilities and compromises necessary in these two very different domains.
You need to make sure the person on the other side is also using a landline, though.
Also, Skype is great when both parties have sufficient bandwidth for whatever codec it opts when it detects a high quality data connection but it simply sucks when it detects poor connection, gets worse than Google Hangouts does.
If you've ever worked in an office with VoIP phones, you know that the level of high quality audio is eerie. It's almost too real to seem like a phone call.
Wouldn't sidetone be generated locally on the device with near-zero latency? There must be some other reason they don't offer it. Maybe to avoid feedback?
tl;dr 1) The choice of digital protocol for PSTN was sub-optimal; 2) landline phones are installed in areas which tend to be quiet and private; 3) the author likes holding a receiver handset, and doesn't consider modern smartphones or earbuds as good replacements.
None of these are really issues.
1) The author invokes the old myth that "digital audio must be sampled, therefore it's lower quality", whilst conveniently ignoring the facts that a) that sampling can be performed to any degree of precision you like and b) those samples can be reproduced exactly at the other end, unlike analogue signals which constantly degrade. Saying that the digital protocol is "permanent" is basically a lie; the fact it's digital, and being consumed by general purpose computers, means it is far easier to upgrade, provide capability-discovery, compress, tunnel through older protocols, etc. than the analogue protocols the author is implicitly comparing to.
2) Mobile phones are mobile. They can be moved to quiet and private locations. The author laments the loss of landline telephones installed in booths, at restaurants, etc. whilst ignoring how much it cost to install and maintain them. If a business/organisation were to spend money on providing customer phone service, it would make more sense to install a cellular repeater than a bank of landlines.
3) Many handsets are available for use with smartphones.
I think the only problems with mobile phones compared to landlines are social. For example, if a landline isn't answered then the owner might be out; if a mobile isn't answered then the owner doesn't like you; if a mobile goes to voicemail faster than normal, then the owner must have pressed the reject button; etc. These just depend on expectations: if you jump through hoops to be available for calls and answer them as quickly as possible, no matter the inconvenience or situation, then you're setting a high bar for yourself. On the other hand, if you only use a phone when it's convenient, and have it switched off or rejecting calls the rest of the time, then nobody can make those social inferences.
> The author invokes the old myth that "digital audio must be sampled, therefore it's lower quality"
Actually, the author made a different argument: That the removal of the higher frequencies in PTSN PCM combined with the low frequency of modern cellphone background noise made for something much harder to understand. As someone quite familiar with the subject, I think this is a very nuanced and compelling argument.
> Saying that the digital protocol is "permanent" is basically a lie;
The author also did a spectacular job dealing with this in an understandable way too, talking about how it's baked into so much of the infrastructure.
> Actually, the author made a different argument: That the removal of the higher frequencies in PTSN PCM combined with the low frequency of modern cellphone background noise made for something much harder to understand.
The author certainly made the case you say, which I don't disagree with. However, they also invoke the "analogue sounds better" myth. Regarding analogue:
> At first, telephone audio was entirely analogue, such that the signal of your voice and your interlocutor’s would be sent directly over the copper wire. The human ear can hear frequencies up to about 20 kHz, but for bandwidth considerations, the channel was restricted to a narrow frequency range called the voice band, between 300 and 3,400 Hz. It was a reasonable choice when the purpose of phones—to transmit and receive normal human speech—was taken into account.
Regarding digital:
> In order to digitally switch calls, the PSTN became subject to sampling, the process of converting a continuous signal to a discrete one... Since the voice band required only 4 kHz of bandwidth, a sampling rate of 8 kHz (that is, 8,000 samples per second) was established by Bell Labs engineers for a voice digitization method.
This choice of frequency is later criticised as too narrow; yet the analogue signal was even narrower! Even ignoring the issues of error correction, etc. this goes completely against the author's claim that "they’ve become much crappier phones... partly because of the nature of the apparatus we used to refer to as the "telephone"...".
>> Saying that the digital protocol is "permanent" is basically a lie; [...it is far easier to upgrade... than the analogue protocols the author is implicitly comparing to.]
>
> The author also did a spectacular job dealing with this in an understandable way too, talking about how it's baked into so much of the infrastructure.
I've added a little more context to that quote: the existing digital infrastructure may be difficult to upgrade, but the previous analogue networks would be harder. Hence it's misleading to make claims like this when arguing that modern phones are "much crappier" than what came before.
In fact, I disagree that digital phone infrastructure is difficult to upgrade. With a few taps on a digital phone (e.g. running Android) a non-technical user can download a program like Skype, over their digital phone network, install it and be making calls over that network in a matter of minutes. A more technical user can use WebRTC to tunnel any protocols they like, at whatever bandwidth they like, recorded with whichever equipment they like, in whatever location they like, all on top of the existing infrastructure. Those are digital phone calls just as much as any standards-body-approved, nationally-implemented, monopoly-owned operator's offerings.
Also, the author doesn't realize (or maybe never noticed?) that modern handsets support wideband codecs, often marketed as "HD Voice," that offer far superior audio than any landline.
On the other hand, there seems to be little interoperability between carriers, at least in the US. Even though all of the major networks support it between their own customers, none of them enable it between carriers.
I reject the use of voice calls (I actually no longer use a phone) because synchronous communications interrupt what you are doing right now, and more importantly because they lack a proof-of-work barrier. Written communications automatically exclude individuals who cannot read and write properly. That is a fantastic time saver, as it keeps your contact list clean. Furthermore, it is trivially easy to implement anti-spam measures, to exclude individuals who still happen to be able to read/write but whose communications are also undesired. Google/gmail even does that automatically. Written communications are merciless in stamping out and excluding the illiterate and other undesired demographics. Therefore, voice calls are not a feature of a phone, but a bug.
A few thoughts from deep inside the VoIP industry:
1) There are "high definition" / wideband codecs that remove the traditional 3.1 KHz PCM bearer channel audio spectrum restriction, along with its logarithmic companding, such as G.722. There is also desk handset hardware that makes them sound really, really good; given both, it's unlike any phone call you've ever had.
2) The challenge is that even in 2016, the PSTN - with its 3.1 KHz bearer channels - is still the common denominator to which nearly all audio must be reduced. So, you only stand a chance of having that kind of call quality on a private/intra-PBX type call.
3) Direct IP peering among VoIP providers hasn't really taken off, although pure-IP signalling across interconnects is becoming increasingly common. This is largely due to institutional inertia and deliberate lobbying on the part of LECs of all kinds who are invested in the incumbent per-minute revenue models and cost structures of the PSTN. (Even though the measured usage aspect is generally hidden from the end-user, the intra-industrial cost structure is not at all flat-rate.)
4) The VoLTE ambitions of mobile operators might change the game a lot as far as call quality, but there is some question as to whether the spectrum is physically there to support HD calls for all, and of course the limiting factor of mobile handsets as described in the article.
5) Residential and consumer users are probably not willing to pay for premium call quality experiences or value-added application-level features, and the VoIP industry is mostly disinterested in selling to them except through wholesale channels. But core business users - folks who still use desk phones, and who are pretty much the exclusive focus of the VoIP industry at this point - do care about this, and there's active work going on to cater to them and make their conference calls and such sound better.
6) WebRTC is overhyped and overrated, but there is one large industry to which it is directly relevant and for which it will make a significant impact: contact centres (call centres, customer service centres, support centres, etc.). Better codecs can be found there also, though somewhat offset by the varying quality of Internet connectivity and intermediate network engineering (that call centres are cheapscates is an understatement).
7) Along the lines of #4, over-the-top applications that support voice calling, such as Skype and WhatsApp, are also increasingly changing the character of the typical Millennial's "phone call", especially outside the USA. However, they are aggressively resisted by mobile carriers for obvious reasons: it commoditises them down into a dumb data pipe and purely price-level competition to provide bandwidth, while some clever Silicon Valley app takes all the value-added / higher-margin premium.
> but there is some question as to whether the spectrum is physically there to support HD calls for all
This surprises me. I assumed that with all the spectral efficiency gains that've been made to support higher data usage, doubling (or more) the bit rate of voice calls wouldn't be a big deal. Now I'm curious what percentage of overall data traffic on a cellular network is voice.
Yeah, but the data usage is all asynchronous stuff, not especially delay-sensitive. It can be shipped in large packets. Voice isn't YouTube; it can't be buffered (much), so it's a hailstorm of a very large number of very small packets[1] that all have to get there in a sufficiently quick, reliable and ordered fashion as to approximate the circuit-switched experience. That changes the physical layer oversubscription economics significantly.
[1] A typical "packetisation duration" in wireline VoIP is 20 ms. That's the duration of audio buffered into a single RTP frame.
Edit: I'm not saying definitively that it's not physically possible to deliver high-bandwidth voice conversations for everyone as table stakes. I'm ultimately a fixed-line guy and not in a sufficient position of knowledge in wireless/RF to know that for sure. I only meant to say that there are known questions about its realism.
Try this just for fun: Call someone right next to you via your cell phones and talk to each other. One finds one can speak several words, maybe even a short but complete sentence before the other person starts hearing any of it. It can become a laughably long latency when experienced so directly.
But this cumulative latency ( be it from queuing delays, voice packet grouping for burst transmission, buffering delays along the many packet switches from here to there, you name it ) comprise engineering trade-offs necessary for optimal radio-level multiple access, efficient packetizing, and economically efficient (packet) transmission.
In other words, I'd postulate we can't (today) afford full duplex, sidetone-included, wideband (5KHz+) sub-20-millisecond latency in cellular telephony. Cellular is a different thing, but surely it's its own kind of magic.
But consider modern office telephony over Ethernet: I don't know the internals of it, but aesthetically, I find it significantly better than the analog telephony I remember from 40 years ago: The sidetone's there, it's full-duplex, speakerphone functionality is absolutely superb, and subjectively I find it sounds good.
I'd guesstimate a three orders of magnitude latency ratio between LAN-based telephony vs cellular telephony seems to me key to the very different engineering possibilities and compromises necessary in these two very different domains.