Myth-busting: Audio Bandwidth vs Data Bandwidth


Few things get me as agitated as the flagrant spreading of disinformation on the part of the plainly ignorant or apathetic. How’s that as an opening line for an argument? Well, it’s a fact.

As mentioned previously, recent weeks have seen some developments in the war against low-def voice. HDVoice service in the UK is getting easier to find, even in the mobile space.

This is very encouraging. Like the adoption of color TV, or music on CD, or video on DVD, the publics exposure to the new technology will create demand where once the established industry players thought there would be none.

In one of my daily visits to Google Reader I found a nice piece on the coming of HDVoice over at TechEye.Net. That article inspired a post on Slashdot. Typical of the Slashdot community there was a long trail of comments. Some of those who commented were reasonably well informed, some with solid telecom experience.

Then there was Rickb928:

Frequency response is not the same thing as bandwidth (though they are directly related), but for telephone a 300-3300Hz response is intelligible and manageable. Doubling it to 6500Hz doesn’t do a whole lot except consume bandwidth and marginally improve intelligibility. If you want fidelity, well, 12,500Hz is a good start. A loty of people never heard the flyback transformer on their old TVs vibrate, but I can hear them loud and clear. That’s 15,750Hz.

Wrong, wrong, wrong…dead wrong! Context is important, and he skips it completely.

For the purposes of conveying the human voice increasing the usable audio passband to 80 Hz – 7 KHz makes a DRAMATIC improvement in call quality. If you can’t hear the difference…well, our hearing suffers through age and abuse.

Don’t just take my word for it. I’ve put a considerable library of audio samples online to prove the point. That material was created in support of my presentation to Astricon 2009, which specifically highlighted the pending availability of the Polycom Siren 7 & Siren 14 codecs in Asterisk. All of the examples were created by passing an uncompressed audio signal through the actual call path. None of it was synthesized using post-production tricks like EQ.

I specifically used various languages to highlight the fact that the improved call quality is most helpful in cross-cultural circumstances. From my recent experience with Tieline I’ve come to appreciate that it also helps when the conversation is happening in a noisy location.

My impression from working with various HDVoice codecs is that the jump from the legacy 300-3400 Hz pass band to 80 Hz -7 KHz is in fact the most substantial improvement. The further jump from 7 KHz to 12-14 KHz in the upper-end delivers considerably less improvement in clarity.

Is 7 KHz the ideal high-end  response limit? No…it’s not. But it’s a good place to start for telephony, where our focus is voice. It’s something that we can do today based upon existing standards that are not encumbered by potentially burdensome patents and royalty schemes. With luck the IETF CODEC working group will arrive at a new standard that combines various best of breed technologies arising out of SILK, CELT, BV32 and SPIRIT IPMR.

David Frankel of ZipDX has often made the point that acting now to implement wideband telephony, even using the elderly G.722 codec, is a far better idea than letting the technophiles argue the merits of their various codecs endlessly while the public continues to suffer with audio standards from the 1930s.

Rickb928 further asserts that wideband audio will consume more bandwidth…which is also dead wrong! It really kills me that this myth hangs on with such tenacity, as witnessed again over at the DSL Reports VoIP Forum just recently.

Let me give you a singular, sound-bite-like thought to drive this myth into the grave:

The entire universe of voice-centric wideband codecs are more bandwidth efficient in encoding voice than the existing PSTN-derived standards.

Even the aged G.722 uses the same bandwidth on- the-wire as it’s narrowband predecessor, G.711. Only those codecs that target full-bandwidth audio, support stereo, and employ data stream redundancy (FEC) offer support for bit-rates higher than presently common on the PSTN.

The simple fact is that all the more modern codecs leverage newer compression schemes that take better advantage of the available data path to provide a higher quality call path in the same or lower bit rates. Some also incorporate packet loss concealment and data stream redundancy to improve the call quality over questionable IP networks. Historically these functions were implemented outside of the codec as it was assumed to be riding a TDM network.

The cellular carriers are especially sensitive to “spectral efficiency.” This is the very reason that they were willing to standardize on AMR-WB and EVRC-WB, both of which bear costly licensing schemes. They see value in ensuring low bit-rates, even as they deploy improved call quality.

He goes on:

To ask for improved sound quality in telephone is to ask for some compromises – fewer conversations over a given link, fewer conversations per cell tower, more Internet bandwidth. I’m pretty sure none of the incumbents will bother, as this ultimately results in increased direct costs, and probably zero increased revenue. Skype, etc., play with the codec and give apparently better results, the emphasis on ‘apparently’. There are some clever audio tricks that will give a more pleasing experience with very little increase in bandwidth. Maybe Android can play with the audio, but I bet Apple could care less. The ILECS, bah!

I certainly agree that we can’t expect innovation from ILECs. Their business model revolves around continuing to wring revenues from the status quo, typically through falsely sustaining the economics of scarcity.

Yet, HDVoice is really gaining ground. The combination of quality IP phones and a hosted IP-PBX allows my employer to enjoy HDVoice calls every day. In so doing we’re also saving money by not paying per minute for inter-office calling. I am not alone in this endeavor. I was recently in an ABC/Disney facility and noted that they had completed their transition to IP phones, and can make HDVoice calls over the company WAN.

In fact, the simple reality that telecom equipment ages and must be replaced suggests that HDVoice will eventually be more widespread. It’s a core feature of the newer equipment. More and more phones, PBX, SBCs, etc are being offered HDVoice capable simply to remain competitive.

So, the legacy of telephony is an old one, and has left us with something that works, but not as well as it could. Just a few more dollars, and you could have better!

This is so very true. Nowhere in the realm of technology have we come to accept a sub-standard quality of service, even as the technology used to provide that service keeps improving and dropping in cost. Sadly, the coming of cellular telephony has in fact trained users to accept ever more degradation in call quality where it’s simply not necessary. We should expect better…no we should DEMAND better.

From the sounds of it Rickb928 comes from a telecom background. I certainly respect that experience. As usual, the devil is in the details. Far too often people make assumptions about better call quality being more taxing on network data rates. Superficially it makes sense, but when you look beneath the surface we find that this need not be the case. There are low bit-rate wideband codecs directly comparable to essentially all of the existing narrowband codecs.

That doesn’t mean that there aren’t some genuine issues involved in rolling out wideband telephony on a broad scale…but data rate shouldn’t be one of the big ones.