by Dave Immer
Editor's Note: In a recent interview with Jeff Berman [April '94 RAP], we got some insight into the latest technology of real-time transmission of high quality audio over telephone lines. We received several calls wanting to know more about this technology. So we contacted Dave Immer and asked him to give us a handy "what is" and "how to" look at ISDN. It's very new and maybe a little pricey right now, but rest assured, this technology will be knocking on your production room door soon enough!
On-Ramp To the Telephone Turnpike
ISDN, or Integrated Services Digital Network, has finally arrived in the US! For years it has been available in UK, Europe and Pacific Rim. ISDN basic-rate service divides a standard telephone line into three digital channels: Two B (or bearer) channels at 64 kbs (kilo-bits per second) each, and one D channel at 16 kbs for a total bit-rate of 144 kbs.
What does this mean to audio production people? It means that broadcasters, studios, talent, producers, composers, agencies, record companies and anyone else involved with audio production can send and receive high bandwidth stereo audio in real time, over inexpensive twisted pair copper telephone wires to and from anywhere in the world where ISDN is available. What's more, different and separate material can be sent in both directions simultaneously. To do this requires a digital audio "codec" (coder/decoder). One can also send digitized audio files (or any type of data, for that matter) over ISDN at a rate 10 to 15 times faster than a POTS (Plain Old Telephone Service) line for roughly twice the cost of a standard call (domestic). International ISDN (data) call charges tend to be 4 to 6 times the cost of POTS.
Obtaining ISDN Service
In areas where ISDN is available, installation costs run about the same as a business phone line, somewhere between $70 to $400, depending on how aggressively the local telco is promoting it. Try to find the person most familiar with ISDN or switched digital services within the small business accounts division of your local telco and give them your telephone number prefix. This will allow them to determine if ISDN is available at the central office that serves your area. If it is, ask for 2B+D basic-rate ISDN service with the lines configured as alternate circuit switched voice and data on one B channel and circuit switched data on the other B channel. You will most likely need to know the "Service Profile ID" (SPID) of your line to configure your terminal adapter. Finally, find out if the ISDN switch at the central office is either an AT&T 5E, or a Northern Telcom DMS-100 and let the vendor who sells you the terminal adapter know this. That's about it as far as ordering the phone service goes.
Using It
To make an ISDN data or codec call, both caller and recipient must have the service. You can send audio files (like Digidesign) from computer to computer using a PC or Nubus expansion card that connects to the ISDN line. Using this approach, it takes about 15-20 minutes to send 1 minute of "uncompressed" 44.1 stereo audio. This is useful if you are doing critical mastering and don't want to suffer the disadvantages of coding and decoding audio (more about this later). Plus, the hardware and software cost is under $2000 on each end. This would also make sense in a "store and forward" type system for spot distribution and such.
But, the "Real-Time" method using audio codecs is what is going to revolutionize audio production. Using one ISDN line (to keep long distance costs down), there are presently three systems out there that sound great. The Musicam ISO/MPEG layer 2 algorithm employed by the CCS CDQ2000 and RE660/661 codecs, and the ISO/MPEG layer 3 algorithm used by the Telos Zephyr, feature full bandwidth stereo audio transmission in both directions with some coding delay (about 0.2 seconds). This would be ideal for recording an announcer, singer or player who needed to hear a cue mix while delivering a performance in the other direction. One advantage of layer 3 is it provides true dual mono audio for complete stereo separation at 128kbs, whereas layer 2 must implement a "joint stereo" mode that sums redundant information between the two stereo channels to achieve full fidelity at 128 kbs. Still, I have not found this to be a problem in using the CDQ2000. With a higher bit-rate (requiring 2 or more ISDN lines), the APT-X algorithm can be used. This algorithm has the advantage of very low latency in its coding process, allowing for performers at each end to "Play live" with each other with negligible delay.
Caution!
There is a caveat to be mentioned regarding the use of any codec. That is, the algorithms perform a process called "perceptual audio coding," or PAC, which reduces the bit rate required to reconstruct the audio waveform at the receiving end of the connection. To do this at 128 kbs, over 90% of the original data must simply be thrown away, never to be seen again. While these codecs do an excellent job of hiding the audio artifacts that result from the coding process, we should all be aware of their weaknesses. For this reason, at 128 kbs it is best to limit the program material to one coding/decoding cycle only, and to perform any EQ, limiting, or other post-processing before the audio gets sent. At higher bit rates you can get away with more code/decode cycles and can post-process more freely. Also bear in mind that as more stages of the production, distribution, and broadcast chain employ these codecs, the more the program material goes through code/decode cycles. This is called "cascading" and is a real hazard. Use codecs with care.
Bottom Lines
Expect to spend between $6,000 and $10,000 per side when you go the real-time codec route. The Telos Zephyr goes for around $6,000 and includes the terminal adapter. The RE660/661 costs around $6500, and the CCS CDQ2000ED has a $7000 price tag. Both the RE and CCS product need an external terminal adapter that will cost about $1100. All ISDN systems require an NT-1 network terminator that performs echo-cancellation on the data lines -- about $225. There are many more MPEG layer 2 systems (CDQ-2000, RE660/661) in use than layer 3 systems (Telos Zephyr). So if you want connectivity with existing users around the world, there is a strong argument to be made for MPEG layer 2. For the higher bit-rate boxes requiring multiple ISDN lines, the APT-X seems to be the current leader in existing users. Dolby also makes an excellent box (AC-2).
Alternatives
Having said all this, there is one other approach you could take if you wanted to obtain these capabilities without dealing with all the details. Value-added networks such as IDB Broadcast offer a "turnkey" service in which they will set you up with the phone lines and rent you the equipment for a monthly fee. This may make sense to some users, and one can easily see the advantages. There are a few problems I have with a network approach. 1. It excludes independents from its list of ISDN users, and in some cases "locks out" indies by using encryption chips in their codecs, thus preventing interconnectivity between otherwise identical equipment. 2. If you become a long term user, the eventual cost of renting surpasses owning the equipment. 3. It prevents and discourages users from empowering themselves with the knowledge of how to obtain and use ISDN, which is, after all, a public telephone service.
Use ISDN For All Your Telephone Services
Finally, the ISDN line can still be used for regular telephone calls if you have a special ISDN telephone. These cost around $400-$500. This way, when you aren't sending audio with the codecs or data with the cards, your ISDN service can still work for you, thus increasing its value further. Also, with the PC or Nubus expansion card mentioned earlier, you can send and receive faxes. My recommendation is to GET ISDN if it's available, and add capabilities as you can afford them. Codec rentals should soon be available. Getting the ISDN service is no more expensive than a standard business line and will enable you to quickly take advantage of its vast potential.
♦
