How ISDN Applies To Modern Musicians

The purpose‑built 'virtual overdub suite' ISDN studio at London's h20 Enterprises.

ISDN is a term that you will have certainly have come across in the pages of SOS over the last few years, but perhaps without really knowing what it does and how it works. Hugh Robjohns explains the relevance of this international digital telecommunications network to the modern musician.

ISDN (the acronym stands for Integrated Systems Digital Network) might sound like one of those purely technical issues that perhaps doesn't really belong in the pages of a music recording magazine, but believe me, it does. And the reason it does is because it has already significantly changed the way in which music is recorded. Any situation in which high‑quality audio needs to be transferred over long distances in real time, perhaps between studios around the world, or just between a radio reporter and his newsroom, is a candidate for employing an ISDN link‑up. A vocalist in a recording studio in the UK can overdub live to a studio in America, or finished album masters can be auditioned and approved at once by record company executives without them having to be present at the mix. Busy producers can supervise sessions remotely whilst still auditioning in full quality, and one band has even managed to complete a 'nationwide tour' of radio stations without ever leaving their studio, giving all their interviews and 'live' performances over ISDN lines.

This is not science‑fiction, it is in use now, and on big‑budget productions where time is money, it is also highly cost‑effective, saving both the schedule disruption and expense of perhaps getting a top vocalist back into one specific studio for what might be just a few minutes worth of overdubs. It is also, of course, both much quicker and safer than airmailing precious session tapes or mixes around the world.

Any situation in which high‑quality audio needs to be transferred over long distances in real time is a candidate for employing an ISDN link‑up.

ISDN is nothing more complicated than a type of digital telephone connection. The 'Digital Network' part of the acronym is self‑explanatory — the entire link between sender and destination, through the whole telecommunications network, must be completely digital, avoiding any of the remaining analogue links (many of the international trunk circuits are still analogue). This enables digital source material, which could be anything from audio or video to computer data, to be simply transferred to the destination without the need for any interim conversions or messing about (except for some ISDN links to parts of America and Canada, which needn't concern us here!). The 'Integrated Services' part of the term arises simply because the system is able to provide more than just basic audio and data transmissions — it also facilitates helpful additional services such as caller identification, call back, call waiting, conferencing and many others features which are increasingly being introduced into the domestic telephone networks as the technology filters downmarket.

In order to use an ISDN link you will need to install a new digital phone line — BT's ISDN 2e system (more on this in a sec) would be an example in the UK (most of the telecom service providers in countries around the world can supply the kind of digital systems described in this article, but since every company has its own marketing names for the various systems, I will stick to describing only BT's services). Usually, it is possible to just reconfigure the existing overhead cable into the building to carry the digital signals so, in fact, only the connection box and the internal telephone wiring may actually need to be changed. Then, once you have access to the digital telephone network, you need a means of converting your data into the correct format for transmission. This is the job of a device called a terminal adaptor, or TA (see box 'The ISDN Installation'). This is your interface to the ISDN system and incorporates the keypad necessary to dial out to your destination.

You may also need to apply some form of data reduction before you will be able to fit a high‑quality audio signal (typically around 700kbits per second for each mono channel) into the available ISDN bandwidth (perhaps 256 or 384kbits/second, depending on how many lines are employed). This is done by a device known as a 'CODEC' (derived from its joint function as both enCOder and DECoder) which, for convenience, usually also incorporates a terminal adaptor (see 'Coding and other Secrets' section, later). Some of the better known audio data‑reduction processes used in this application are Dolby Fax, APT x100 and Musicam, and all are now capable of producing remarkably high‑quality audio after decoding. Of course, in an ideal world, we wouldn't be using any form of data reduction, but you have to remember that ISDN was originally developed with the requirements of the small business user in mind, rather than the audio community — high‑quality audio usage, as we will explain later, is made possible by clever 'fixes', such as using multiple lines simultaneously.

How ISDN Works

Multiple ISDN lines are often used to provide a greater data bandwidth (three aggregated ISDN 2e circuits provides 384kbit/S). However, the circuits are not guaranteed to travel the same route and so the terminal adaptors have to be able to compensate for substantially different transmission delays. In this example, one ISDN circuit travels under the Atlantic, another via Iceland, Greenland and Canada using microwave links, and the third by satellite.

A basic ISDN system (like BT's ISDN 2e) using the existing overhead wires provides the subscriber with three multiplexed data circuits — two 'bearer' or 'B‑channels' and a signalling circuit called the 'D‑channel'. The B‑channels are 64kbits/S bi‑directional data streams which carry the actual data, be it audio, video or computer data. The 16kbit/S D‑channel is simply used to initiate and control the communication and is not normally made available for public use.

Although Europeans defined the 64kbit/S 'generic standard', some countries had their own subtly different versions with incompatible signalling and user data‑bit allocations. Eventually, however, sense has prevailed and the unified (ISDN 2e) standard was agreed about two years ago with all European Community countries conforming. All new ISDN 2 and Highway installations are of the ISDN 2e type.

There are actually no rules as to how the data capacity of an ISDN circuit can be used, so in an office environment the user could decide to allocate one of the B‑channels to a conventional telephone service, whilst the other is employed as a dedicated high‑speed digital fax or Internet connection. An ISDN 2e installation can be allocated up to 10 telephone numbers to address specific digital telephones, fax machines, or whatever, but as there are just two B‑channels, only two devices can be used at the same time. Alternatively, the two B‑channels can be used together ('aggregated') to form a single 128kbit/S channel, but with audio applications in mind, this still doesn't come near to the 700kbit/S of a single uncompressed audio channel — typically either four or six B‑channels will be used (combined with data reduction on the source signal), providing 256 or 384kbits/S capacity.

With suitable multiple‑line equipment, additional ISDN 2e lines can be used to increase the data rate even further. BT offer a 'Primary Rate' service called ISDN 30 which can be configured for between 8 and 30 standard ISDN circuits — the latter providing a 2Mbit/S link more than sufficient for uncompressed stereo audio. But aggregating multiple ISDN lines is where the fun really starts because ISDN was not really designed to be used in this way. The problem is that the separate circuits being used simultaneously can be directed to the destination over different routes, thereby incurring substantial timing differences. There is actually no way to guarantee that each line will follow the same path — in a six‑B‑channel call to Los Angeles, you might end up with the first pair routed via a cable under the Atlantic, the second pair via a direct satellite hop, and the third via a complex sea/land/satellite route via Iceland, Newfoundland, and Canada! Terminal adaptors and audio codecs therefore have to be able to recognise the path delays and then resynchronise data from the various ISDN circuits. This inevitably results in the overall signal path being delayed by the time taken by the slowest circuit.

In practice, most codecs have very elegant and sophisticated approaches to handling multi‑line calls and it is rare to experience insurmountable problems. If a line drops out for some reason, the better codecs will automatically reconfigure the data reduction and reallocate data around the remaining lines without a noticeable break in the transmission. Some systems will even automatically re‑dial the lost number and when the circuit is re‑established, revert to the original configuration!

ISDN In Practice At h2o

h2o's Andy Hilton overseeing the company's first ISDN session in their new premises while the studio was still under construction!

One of the leading exponents of the use of ISDN technology in professional audio applications in the UK is the south London‑based h2o organisation, founded by Andy and Robin Hilton. They claim the h2o facility to be the first purpose‑built ISDN 'virtual overdub' suite in the world. The operation consists of an overdub booth and control suite equipped with 14 B‑channels of ISDN, together with all three principal types of codec. The booth is fully isolated and floating from the rest of the building's structure and the control room is equipped with a Yamaha 02R digital console, interfaced digitally with a wide range of DAT, CD and modular digital multitrack (ADAT and DTRS) recorders, along with the assortment of ISDN codecs. Apogee converters handle analogue sources from the selection of professional microphones and quality outboard equipment, whilst Steinberg's Cubase VST/24 handles hard disk audio editing and sequencing. The audio codecs and terminal adaptors are housed in an adjoining machine room and controlled remotely via a couple of PCs near the console. These provide ISDN phone number databases, automated dialling, system monitoring, and various other housekeeping functions.

The bulk of h2o's work can be categorised as either audio transfers, remote overdubbing, or 'bridging' — all of which are available 24 hours a day (essential to overcome time zone differences with clients in all corners of the globe). The term 'bridging' means linking between two or more separate ISDN lines, either to overcome incompatible audio codec standards, or to provide privacy and security, for example by isolating an artist being interviewed at home from a broadcaster.

Where incompatible codecs are used, or to keep the source ISDN phone numbers secret from an end user, ISDN bridging is necessary. A third party provides the interface between the two users, converting between codec formats and isolating them from each other. Depending on the codecs involved, the bridging may be performed digitally or via trusty old analogue!

Although there is no such thing as 'a normal day' at h2o, the company will typically handle between three and five ISDN sessions a day, with work from New York tending to fill in the early evening and Los Angeles in the small hours of the morning. The company also hires out both equipment and its ISDN expertise to other studios and individuals. Many studios can justify installing permanent ISDN lines, but buying all three commonly used audio codecs is prohibitively expensive unless they are in regular use, so it pays to hire units when they are needed — Air Studios and Abbey Road both take this approach.

h2o will typically handle between three and five ISDN sessions a day.

The transfer side of h2o's business typically involves sending or receiving audition copies of album mixes, or approval copies of radio commercials. This is a particularly effective way of working when different time zones are involved. For example, if an artist is mixing an album in New York but the recording company executives are in London, using ISDN to transfer some full‑quality alternative mixes overnight allows the record company execs to listen and comment before the next day's work commences back in the States — infinitely better than playing material over analogue telephone lines! In the example above, getting virtually instant feedback on a mix project allows it to be completed much more efficiently and quickly, and the time and money saved on returning to mixdown suites and resetting the console and outboard to remix a track far outweigh the modest expense of real‑time ISDN transfers.

h2o often find themselves transferring orchestral film soundtracks recorded in the big London studios to film dubbing houses in Los Angeles — sometimes purely for approval, but occasionally to be used on the final soundtrack when the deadline has been reached!

The overdubbing side of the business started with voice‑over work for overseas radio and television commercials, for example where a 'British voice' was needed, or where a foreign accent was required in a UK advertisement. After all, if you want a German‑speaking voice‑over artist, you are probably going to find the best in Germany — and when the job itself probably only takes 10 minutes, ISDN offers a much cheaper alternative to air fares! Charges at h2o, for example, are £125 per 30‑minute session during the day, rising to £185 at night. Call charges are extra, as the rates vary enormously depending upon which country is involved. Apparently Japan and Australia are the most expensive countries for ISDN calls at present, but a six B‑channel call to America is actually cheaper than the same call to most European destinations!

Working Methods

Far right: h2o's reception area.

An extension to this kind of work is the musical overdub, hence h2o's claim of providing a 'booth to the world'. Any suitable ISDN‑equipped studio, anywhere in the world can make use of h2o's booth to record an artist in the UK, thus saving on time and travel. Although ISDN is able to transfer audio data in real time (ie. a three‑minute song will take three minutes to transfer), both the coding process and transmission distances involved mean that there is an operationally significant delay (possibly up to half a second) between the audio being sent and arriving at its destination. There are two primary methods of working around this: the first is to receive the necessary tracks and submixes via ISDN from the overseas studio and then, using SMPTE timecode for synchronisation, rebuild enough of a replica of the original multitrack tape to allow the artist to overdub the required parts. The producer in the overseas studio can still supervise the session, with full quality monitoring and communication via ISDN. The overdubs are recorded on a multitrack in the local studio, and the tapes sent back to the overseas facility for laying back the new parts on to the original project multitrack tapes.

With the hectic schedules of some international artists, this can be a very pragmatic way of working, taking advantage of snippets of spare time. For example, when Mariah Carey needed to do some repair work to a vocal track on her duet with Whitney Houston for the title song for Spielberg's Prince of Egypt animated feature, the whole session was set up at h2o within three hours of receiving the request from the producer and saved a great deal of expense and time compared with trying to get the singer back to the US for a session in an American studio. Another particularly spectacular example of a last‑minute ISDN transfer saving the day was when the multitrack tape required for a recording session in the UK with Placido Domingo was inadvertently left behind in the USA! Transferring 12 tracks of submixes against timecode via ISDN allowed a temporary replacement multitrack tape to be built up so that the session was still able to go ahead exactly as planned.

...when Mariah Carey needed to do some repair work to a vocal track on her duet with Whitney Houston for the title song for Spielberg's Prince Of Egypt animated feature, the whole session was set up at h2o within three hours of receiving the request from the producer...

The second approach to ISDN overdubbing is to send the actual monitor mix directly from the overseas studio, and simultaneously record the performance back on to the original master tape also via ISDN (a backup recording should always be made in the local studio too). The main problem with this technique, of course, is that the overdubbed audio will arrive back at the remote studio out of time with the rest of the track. However, the offset will be consistent, for once an ISDN connection has actually been established, the delay time remains fixed, so it is actually quite simple to slip tracks back into synchronisation after recording.

Coding & Other Secrets

The recording industry at large is still somewhat apprehensive about audio data reduction — often as much because of fear of the unknown as any justified complaint over quality. Full‑bandwidth, complex audio signals such as an orchestra or complete band mix can suffer perceptible degradation when subjected to heavy data reduction, but in an application such as the 'virtual overdub booth' where only a single instrument or voice is involved (with a very high degree of redundancy in the signal), there is a negligible quality change. Indeed, some of the larger motion picture companies seem perfectly happy to use orchestral soundtracks transferred over ISDN for film masters, and there have even been occasions when album tracks have been mastered directly from ISDN transfers.

An alternative approach to the virtual booth avoids the need to transfer sections of the original multitrack in advance of the overdubbing session. A monitor mix of the original multitrack is played to the artist in the remote studio via ISDN (1). The artist performs to this backing track and is recorded locally for safety (2) whilst also being recorded in the source studio. The delay in the ISDN round trip is compensated in the monitoring by an equivalent delay in the multitrack mix (3).There are three commonly used types of ISDN audio codecs: Dolby Fax, APT x100, and Musicam (MPEG). The first of these tends to be the domain of the movie houses and high‑end recording studios, with only around 300 systems worldwide. The hardware has always been expensive and is no longer marketed directly by Dolby, but it is an excellent sounding system preferred by many top American studios.

Based around the AC2 data‑reduction algorithm (a forerunner to the more widely known AC3), Dolby's system is unusual in that it does not have an integral terminal adaptor. Instead, Dolby have concentrated on providing the best possible audio codec, and have left the user to sort out the line termination and call handling separately, via any industry‑standard TA.

Dolby Fax requires four B‑channels to operate at a fixed rate of 256kbits/S (five B‑channels are needed in countries operating a switched‑56 system), and has the ability to initiate recording on a suitable remote DAT machine (hence the idea of an 'audio fax machine'). Another important feature of the Dolby system is that units can be interlinked to accommodate multi‑channel audio — for example, five‑channel audio for surround mixes. Apparently there are only eight installations in the world equipped to deal with multi‑channel Dolby Fax (and h2o are the only facility in Europe).

APT (Advanced Processing Technology — a company owned by major professional mixing console manufacturer Solid State Logic) developed a data reduction system called x100. This is widely used by broadcasters, with many independent radio stations (Classic FM, for example) relying upon it, and has also been adopted for the sound coding on dts (Digital Theatre Systems) feature film prints. The x100 coding format provides the shortest processing delay of all the common data‑reduction systems (typically tens of milliseconds instead of hundreds), and offers the highest bandwidth, with up to six B‑channels. The ISDN units can also convey SMPTE timecode via the D‑channel, thus enabling synchronisation without having to sacrifice an audio channel.

The Musicam format employs the MPEG (Motion Picture Experts Group) family of coding algorithms and is again widely used by broadcasters. This extremely flexible system requires anything from one to six ISDN B‑channels (depending on the desired audio quality) and varies the degree of data reduction accordingly. A number of manufacturers make Musicam‑compatible codecs, with CCS being one of the most popular. Since MPEG employs a very complex form of perceptual coding data reduction, it tends to involve relatively long processing delays — typically in the order of a few hundred milliseconds. It is impossible to specify this precisely, as the delay depends on the degree of data reduction required.

One thing to bear in mind when working with ISDN is that any audio signal processing that is necessary (such as EQ or compression) is best done, if possible, before the signal is subjected to coding and therefore possibly data‑reduced, since any processing after this point may not have the full signal to work on and, subjectively, rarely seems to work as well.

Any ISDN bureau has to be equipped to handle audio data from all three main codec types, but just having the equipment is not enough — there are, at present, still a number of traps and 'secrets' waiting to trip up the ISDN novice. For example, the American ISDN network has a peculiarity in its design which makes it difficult to reliably use more than two B‑channels simultaneously for a single transmission, as the network timing can vary to such an extent that many codecs can not resynchronise data received on separate lines. Another 'strange‑but‑true' example is that calls to ISDN destinations in Madrid suffer occasional random dropouts because their Network Switches occasionally close down lines for no obvious reason! A further little‑known aspect of international calls can easily catch out novice ISDN users. Normally, an international call from the UK is signalled by dialling two zeros before the country code. However, many of the international trunk circuits are still analogue and if an ISDN call happens to be routed over one of them, the data becomes hopelessly garbled. Fortunately, dialling three zeros instead of two at the start of the number instructs exchanges en‑route to find a completely digital path — not many people know that!

Another common reason for problems with ISDN is when a call is routed via a private telecoms supplier — a cable company, for example. One technique these companies use to reduce their operating costs is to aggregate calls, squeezing as many as possible onto a fixed‑rate digital circuit, thereby ensuring it operates at maximum capacity. However, the constantly changing call loading often introduces varying delays in the traffic and although this has little noticeable effect on normal voice communications, it can completely trash the timing of multi‑channel ISDN data, resulting in more garbled transmissions. By contrast, the BT digital network is engineered to be absolutely rock‑solid in its timing which, although making it rather more expensive, provides far better reliability. Also BT tends to be far more supportive and understanding of its ISDN user base than other telecoms companies because it has greater experience and knowledge of the technology.

The Future

ISDN is already widely used in the commercial world and is increasingly making an impact on the professional audio industry. As the national and global telecommunications infrastructure continues to be upgraded and digitised, analogue telephone systems become increasingly difficult and expensive to support and, in the not‑too‑distant future, ISDN will replace them completely. Although currently more expensive than analogue installations, the cost of ISDN continues to fall and anyone with a requirement for fast, high‑capacity data links should certainly investigate the practicalities and financial implications of either upgrading now, or opting for something like the BT Highway system which combines both analogue and digital technologies.

Although it can still be a minefield for the inexperienced — the current crop of terminal adaptors and audio codecs are not exactly user‑friendly — ISDN is actually a wonderfully reliable system when you know all the tips and tricks. It may appear to be largely the province of a number of specialists today, but as the user‑base grows, manufacturers will inevitably improve the ergonomics of their equipment, and there is no doubting that ISDN will eventually become an everyday technology, available certainly in every professional studio and probably even every project studio too.