The PSTN (Public Switched Telephone Network) has been used for decades to connect a remote reporter to the studio, and many products have sought to make the process simpler and the audio quality better and overcome the limitations of a two-wire system.

Telephone couplers simply provided a method of attaching audio equipment to a telephone line while protecting each from the other.

Telephone hybrids offered an acceptable level of separation between incoming and outgoing audio and allowed professional (balanced 4-wire) audio equipment to connect directly to the network.

Frequency extenders adjusted the frequency range of audio as it went into the network and came out at the far end. The adjustment allowed the resonant lower frequencies to fit into the narrow bandwidth of conventional telephone communications, losing only a small fraction of the higher frequencies as a result. More sophisticated extenders provided a broader range, utilising two or three telephone lines. Still in use worldwide, frequency extenders are robust products that perform their task adequately on quite poor lines.

Now, telephone codecs offer very high quality audio, in both directions, over a single line. These convert analogue audio into data, use processors to reduce the volume of data without losing much audio quality, and then use a built-in modem to send this data via the line. At the far end, an identical device reverses the process. The delay caused by audio processing is relatively short and most products are suitable for live links.

Currently, the telephone codec offers the best audio quality obtainable on a phone line. It is possible to provide a 15kHz audio feed to the studio, with a 15kHz return audio feed for a local PA system, with very short overall delay suitable for live two-way broadcasting, using only an average telephone line for the link.

Given the widespread accessibility to phone lines, it is possible to make a connection from almost any inhabited place in the world. Some phone lines will be too poor to make the connection worthwhile, but in general good quality audio can be sent and received by telephone codecs.

For an extremely good example of a telephone codec, have a look at the Vector on the Audio page.

ISDN simply takes the current telephone network and extends the digital capabilities from the exchange to the user. Three types of service are available: voice (basically the same as conventional telephones); asynchronous data and synchronous data.

The two data services are used to transmit high-quality audio over ISDN. Asynchronous data is used when sending audio from/to devices such as computers, as transmission can be paused while one or other device is busy doing something else. Typically the duration of the call is longer than the audio content, rendering asynchronous calls unsuitable for live transmissions.. For wideband stereo the call may take several times the length of the audio.

Synchronous data is always transmitted in real time, albeit with a short delay for the processing. An ISDN codec processes digital audio to reduce the volume of data to manageable proportions, and sends this data via a terminal adapter (often built into the codec) over ISDN. ISDN codecs are usually full-duplex, sending and receiving audio simultaneously.

the ISDN service is presented in 64kbit/second channels, called bearers, and several bearers can be combined to provide a greater throughput. This process may be performed by the terminal adapters at each end of the line, or by the codecs themselves.

Various different data reduction processes are used, primarily trading speed of processing against audio bandwidth. Codecs that reduce audio data by a factor of four are generally very fast, while those which reduce data by a factor of twelve or more can be too slow for interactive work. In many cases, the latter use the return audio channel simply for communications.

Current digital mobile networks offer asynchronous data links at very low bitrates, hardly suitable for broadcast applications. The next generation (third generation or UMTS) of mobile services promise data links of up to 2Mbits/second, offering the possibility of broadcast audio links from anywhere covered by a cellular network.

Network providers hope to have services in the UK in 2002, and many European countries will not be far behind. Legislation will support roaming, overcoming the short-term problem of limited geographical coverage in the early days.

Services have yet to be agreed, data channels as wide as 2Mbits/second may prove expensive, and the asynchronous system may not be ideal for broadcast services, but the freedom offered by being able to just dial and connect without searching for a suitable wall-socket is an extremely valuable resource.

Satellite data services provide high-speed digital data links from anywhere in the world, offering a service similar to ISDN without the need for an ISDN socket.

Satellite services such as Inmarsat-B provide a 64kbits/second synchronous data link similar to ISDN, and many ISDN codecs may be freely used with these services. The typical concerns are that of power, size and weight. Even when a satellite uplink is described as portable, it is still a sizeable thing and any additional weight must be carefully considered.

High costs, both capital and call charges, tend to limit the use of satellite phones to newsgathering in areas without a basic communications infrastructure.

Internet Protocol is the process by which the packets of data which make up web pages and files are sent from one address to another. This is not strictly a dial-up circuit, as packets of data can be sent by varied routes, arriving in any order at the destination.

Using the internet to establish a broadcast link to the studio has typically meant sending an audio file as an attachment to an e-mail. As with PSTN and ISDN services, the audio information is often reduced in volume to provide, in this case, shorter messages. Audio processing can be achieved by a program running on a PC or by a dedicated hardware device with a PC-like interface to the internet.

Live audio links are limited by the data bandwidth available along the route. In general the user has little control over this at any point beyond the short hop to the service provider, and no control over the route the data follows. Often, routes are determined by current usage and by traffic agreements between service providers. For example, current VoIP services (internet "telephones") are aimed squarely at cost saving rather than audio quality.

Access to the internet is getting faster at every stage of a packet’s journey, and service providers are starting to offer service level agreements for guaranteed transmission of data by a particular route or within a particular time. While these have to be arranged in advance they will be of limited use to the broadcaster, but the opportunity for live, duplex audio over the internet is clearly increasing.

A digital audio codec (COder/DECoder) performs two functions: it turns analogue audio into data, then it reduces the volume of data to manageable proportions, reversing these processes for incoming data. The data reduction is performed according to one of several algorithms, the same algorithm being used for both the encoding and decoding processes. This can result in anything from a fourfold reduction in data to sixty-fold or more, allowing high-quality broad bandwidth audio to be sent over a narrow data channel.

The algorithm used in a codec renders the device suitable for a particular application and a particular transmission medium. Very high compression applied to speech is unnecessary when an ISDN circuit is available, just as low compression applied to music is of little use when using a telephone circuit. Consequently, different algorithms have been developed for each application. 

ADPCM is a low-compression, low-loss technique most often used in ISDN codecs for voice, but also available for high fidelity music transmission on multiple ISDN bearers. Examples are G.722 (an ITU standard) and apt-X 100 (a proprietary algorithm from the company APT).

CELP is a high-compression algorithm developed expressly for voice transmission on narrow-bandwidth channels such as PSTN. Accurate reproduction is available within the normal dynamic and frequency range of speech, although extraneous sounds are reproduced poorly. 

Psychoacoustic algorithms rely on the ear's inability to identify a quiet sound when masked by a louder sound at a close frequency, and they reduce the amount of data by a considerably greater factor than ADPCM. Most stereo music codecs use this type of algorithm, as very satisfactory results can be obtained with relatively low data rates. The delay is considerably longer than ADPCM or CELP, but music links do not generally require the same two-way contribution of speech The most common algorithms are ISO/MPEG layers II and III.