When is enough quality, enough quality?

When the MP3 format first appeared there were some who dismissed it because it was a step backwards in quality. This is true. Many MP3s are encoded at 128Kb/s at which certain slight differences can be discerned on a good pair of headphones.

Obviously, these people were wrong. The slight loss in quality was a small price to pay for the massive savings in file size. Taking a 30Mb WAV file and turning it into a 4Mb MP3 made it reasonable to move music across dial-up connections, and a breeze to move across higher-speed networks.

In the case of MP3s, the quality was enough. As storage and bandwidth becomes more available and affordable, this quality gap between CDs and MP3 will close (though higher-quality Mp3 encoding, and through new file formats).

What originally made me wonder about when enough quality is enough quality was a local phone call with a friend. We were having a quiet conversation and I listened to my friend’s voice through the tinny phone receiver speaker. It occurred to me that I had all of the equipment in front of me to do much higher quality audio communication. My PC has a 16-bit sound card and a net connection that can easily stream great sounding music. I have a great microphone too.

I want higher quality telephone audio. Why has our technology stopped at the current level? Perhaps it is a limitation of the infrastructure, but I doubt this (I’m downloading MP3s at 300Kb/s on my DSL connection on the same phone line). It’s certainly not the speaker and microphone technology that is limiting quality.

So, it’s not the technology, it’s us – the customers. We must not care enough to demand higher quality or pay for higher quality. Apparently, we are not willing to pay for it.

Note to telcos: I will pay for it (I’d gladly pay $100 for a new handset/receiver that gave higher fidelity audio).

HDTV is another example of meagre customer demand for higher quality. There are loads of factors that I don’t claim to understand (or even know about) that have stunted the adoption of HDTV. However, it seems to me that people don’t seem to care.

I’ve seen HDTV at the local FutureShop. It looks great – obviously a far greater picture than plain-old NTSC. Joe public, with his 19” TV watching Good Morning Regis & Stupid doesn’t need anymore quality. Regis looks just fine with XXX lines.

48 bit color? Not unless you are scanning for the cover of National Geographic. 32 bit audio – does it even exist? When is enough quality, enough quality?


12 thoughts on “When is enough quality, enough quality?

  1. Looking at my own experiences as a *consumer*, and how I feel about these experiences (which drive [not just] my buying habits), I see this lack of quality in consumer products as not the result of my unwillingness to pay for a better quality *.*, but rather a result of a clear distrust of the suppliers. I have been lied to and mislead so often by marketeers, I now find I default to utter disbelief in all advertising. I have been too often dissapointed in “quality” products I’ve bought in the past, so now look for cheap ‘disposable’ goods, for the most part.

    Perhaps, if we can remove the crooks from marketing and business in general, we will have quality products worth buying provided by businesses worth supporting.

  2. I think there is another related factor which is the efficency of performance. So much of technology is a chase for a standard, a resting place of quality where the performance of the technology is generally acceptable. Leading the chase are accountants seeing lower per item cost and techices seeking the universal standard. The resulting audio/visual experience may be lacking but that is secondary to the degree of universality of the system. It is clear that MP3 quality is worse than CD just as CD is worse than a clean lp just as a clean lp is worse than live performance. Blame in part the geeks and the need for general technical standards.

  3. The quality of telephones is actually limited by infrastructure, but not in the way that you’re thinking. Various different parts of the network are very highly coupled and assumptions about how the audio works are scattered throughout numerous parts of the system. The only way to improve the quality of the network would be to rip it all up and start again.

    A few months ago I read a story about how a project to improve audio quality had to be abandoned because even making the slightest tweaks tended to break devices that depended on assumptions about how the audio would work. I can’t find it now, but I’m sure it would have been on either http://satn.org or http://frankston.com .

  4. As mentioned above, only a small increase in quality (at the low frequency end) can be accomplished with a better telephone. High frequencies are limited by filters in the line cards at the central office to below 4000 Hz. This is necessary because the voice is sampled 8,000 times per second with each sample generating 8 bits (the 2:1 ratio is called the Nyquist rate or Shannon’s sampling theorem). The result is that 64,000 bits per second are generated in each direction for a phone conversation.

    The entire North American infrastructure is synchronized to a multiple of this 64,000 bps rate. For example, 24 signals are combined to give the T1 rate of 1.544 Mbps (which includes some framing overhead). 96 signals are combined to give the T2 rate of 6.312 Mbps. 4032 signals give the T4 rate of 274.176 bps. Major changes, perhaps by doubling the rate per voice signal, would be needed for the current telephone system to provide wider frequency response. This would cost thousands of dollars per subscriber.

  5. I’ve seen ads of tv for internet phones. Do these provide higher quality sound? I think the ad focuses on it saving companies tons of money.

    How about phone service over cable, or does this have to use the same infrastructure stated above to connect the call?

  6. I saw a wholesale replacement of a telephone system in 1991 when I worked on the Baltic coast of Poland. I met a bunch of Canadians who were in the middle of replacing everything with the best fibre and other equipment Bell had. I suspect it was really just at best an advance to the point the slavic sounding “noworryz” describes above – but this was a leap given the way you answered the phone then was to pick up after a ring and scream “TAK SUHAM” (“yes, i am here”) over and over until the connection took.

  7. I’m curious about how cellphone networks are affected by quality. Is it simplier to upgrade them? Seeing as how many third world countries are forgoing the expensive, land based infrastructure of land lines and going completely celluar.

  8. colin and noworryz both put the finger on the thing – the quality is deliberately limited by the telcos. This is because they put filters to remove frequencies about 4kHz, and do various other tricks, which make a lot of sense if you consider the history.

    This doesn’t mean that you can’t put higher quality speech down a phone line, however, but it could only be done if encoded in such a way that the signal is encoded within a 4kHz bandwidth. So, if you are clever, you might be able to work out an encoding which will work with a modem, but then you’d have to have a decoder at the far end too. Some of the techniques which are used in cell phones (GSM etc) which use code books for voice encoding should be capable of realising better quality over standard telephone lines – but since they are optimised for European/American English speaking voices they may do very poorly on other sources – music for example – for which the telephone system was not designed or intended.

    With the massive increase in backbone network bandwidth which is happening every year (doubles every year) due to the deployment of optical fibre – with spare capacity (dark fibre) and DWDM (dense wave division multiplexing), there should in the future be no real difficulty in providing very much higher quality for telephone services. This may not happen, though, as it appears that standard telephone is adequate for most purposes. Users who want something better may look to other methods for delivery – perhaps Voip or other forms of networked audio.

  9. When I bought my first cellphone and signed up with Bell Mobility in Ontario my service was terrible. I could never get good reception, and dropped calls were normal. Then I called customer service about 5 times to complain. Each time the service got a little better. That’s when I realised that the cell companies (well, Bell at least) use tiered service quality depending on how long you have been with them. The earlier customers got the best service while the new signups got the shaft.

    Dastardly, I say.

    May I also add that Dolby surround sound THX Ultra2 telephone calling is way overdue, along with high definition faxing.

  10. FYI, the frequency range for the phone system is 300-3300Hz (giving a 3kHz BW), although as prev. mentioned it is possible to get more information on the line by, in a sense, multiplying this range a bunch of times along the frequency spectrum of the line (kind of hard to explain w/o a graph). This is how DSL works.

  11. When you talk about sound “quality” you are really only talking about the amount of numbers streaming back and forth between your telephone and the other person’s telephone. To me, quality means a product that gets the job done, is easy to use and doesn’t break. Whether I can hear a perfect representation of the other person’s voice doesn’t really bother me.

  12. Just looking at old postings, and came across this one again. In response to smartin’s posting, I would say that the trick he refers to is really one of expanding the bandwidth. It’s brute force really, since signals up to and perhaps even over 1 MHz can be rammed down a phone line if there’s enough gain. I think we are talking about using relatively high power at the transmitter, and using a lot of gain at the receiver – maybe 100dB or more, though it’s frequency and distance dependent. The point is that the telephone wire does have a bandwidth of around 4kHz, but it’s only approximate, and it does tail off as the frequencies rise. The transmitter/receiver combination can, to some extent compensate for this, and since the data is transferred in the digital domain, all that is required is that sufficient number of bits are transferred accurately enough. Some errors can of course be tolerated, using error correction techniques.

    This is also why DSL only works at high data rates over
    links which are close to the distribution point. Some DSL systems use rate adaptation, to provide reasonably high, but not maximum, data rates out to further locations.

Comments are closed.