A great save
Oct 1, 2003 12:00 PM, Chriss Scherer, editor
The IBOC system, branded HD Radio by its developer, Ibiquity Digital, got a shot in the arm in August when a new audio encoding algorithm tailored for the terrestrial digital system was introduced. The new algorithm, called HDC, was developed jointly with Coding Technologies. Coding Technologies is the creator of Spectral Band Replication (SBR), which enables improved audio quality at extremely low bit-rates. You may know Coding Technologies as the creator of the �Plus� in AAC Plus, the AAC encoding enhancement.
The announcement came at an opportune time. The NRSC had suspended its evaluation of the IBOC system because of the audio quality. It has since reactivated its work. With the ongoing delays with a widespread IBOC rollout, rumors abounded about the future of an IBOC system. For now, the work appears to be moving forward again.
The coding issue has always been a sore spot for IBOC. Lucent's PAC algorithm was showcased as the system to use, but it had always been under fire. It was part of the IBOC mix from the merger between USA Digital Radio and Lucent Digital Radio in 2000. Previously, USADR had used AAC. AAC was also used in the tests that were in the reports submitted to the NRSC. PAC had no track record and no submitted test data to back the claim that it was ready. However, broadcasters knew the score and made their voices heard.
According to Ibiquity, HDC had been under development for some time, which was a well-kept secret. The payoff looks good, as reports from listening demos show that HDC makes the grade overall.
Some demo CDs were made available to NPR personnel for evaluation. I also received a demo CD and was able to play it in several environments, which allowed me to explore the nuances in the system.
The demo CD contains 10 audio source samples. Each source is presented in seven formats: the original source, FM analog, HDC at 96kb/s, HDC at 64kb/s, AM analog, HDC at 36kb/s and HDC at 20kb/s. Two of the samples are spoken voice (male and female), one is classical, one is a typical promo/commercial and the remaining six are various styles of rock and pop.
The comparison between the source material and FM and AM analog broadcast quality provides a baseline reference to current technology. The analog samples are listenable and have the conventional sound of our current technology. In my own listening tests, which were not blind, I was able to hear differences between some of the encoding schemes, but the differences were not troublesome.
Obviously, the highest bit-rate (96kb/s) sounds the best and most like the original. Even with careful listening between the two, it was difficult to tell the difference if I could hear one at all. As I switched to the lower bit-rates, some subtle differences began to appear.
At 64kb/s, I could sometimes hear a slightly metallic edge to the audio, particularly in the high frequencies. The 36kb/s samples had a reduced stereo separation and the high frequencies did not have the same sparkle (slight high-frequency roll off) as the original. At 20kb/s, there was no stereo separation and the high-frequency roll off was more pronounced, but it was still listenable and a marked improvement over current mono AM.
All the differences I detected were subtle. Unless a listener was able to directly compare the original source to the encoded version, I don't think that he would be able to tell a difference. It appears that the Ibiquity team has found a good solution to the audio encoding problem, and I congratulate them on their accomplishment. Now that this major obstacle is out of the way, work can continue in the other areas.
Send comments to: E-mail:firstname.lastname@example.orgFax:913-967-1905