
Overcoming the challenges of wireless audio distribution
More unexpectedly (to those who don’t fall into this category) there is a demand for sight-impaired cinema-goers to hear a narrative track to help them follow some details of the plot, and this content is being included in new films today. The cost and inconvenience of providing wiring to every possible user position is such that wireless technologies are being almost universally adopted for these sound feeds.
Several popular technologies around
Induction loops around auditoria are fed from a current amplifier, which is fed from the audio amplification system, or from special microphones. All hearing aids include a pick-up coil that can be selected as alternative to the internal microphone. Loops are relatively cheap because the user provides half the system, but they are prone to interference — the 50 or 60 Hz mains power and its harmonics fall well within the pass band — and they are limited to a single sound channel.
Infra-red systems comprise a number of LED emitter units which are fed either directly with audio (again with interference concerns) or with a modulated carrier, which can deliver multiple audio channels for stereo or for user selection, or both. Both analogue — pulse frequency modulated — and digital QPSK systems are used, generally with carriers in the range 2-6 MHz. Multiple channels are possible for conference and public systems. Infra-red equipment is fairly low-cost, with the disadvantage of needing a direct line of sight, or at least a good reflection, between the emitter and each receiver. Reflective surfaces are unfortunately undesirable in auditoria for many reasons. This means that multiple emitters have to be installed, making the total cost of a system quite high. Retro-fitting infra-red systems can be an especially tricky and expensive operation.
Radio systems classically rely on licence-free pieces of spectrum (bits around 30 MHz, bits around 174 MHz, 863-865 MHz, 1795‑1800 MHz are allocated in most of Europe) or on licensed spectrum, commonly in unused television channels. Radio has the advantage of not needing a line of sight, but many of these VHF and UHF systems are under pressure either from governments who want to sell the spectrum, or by potential interference from other systems — like radio microphones — which share the same allocations.
Wireless audio distribution often takes place in dynamic, noisy environments, often close to many other competing wireless signals. How can these issues be overcome to deliver a better wireless audio experience in public venues? What is needed is a radio system with a reasonably generous licence-free allocation of spectrum, very low component cost and digital modulation to avoid interference.
Digital enhanced cordless telecommunication – DECT
Those parameters almost exactly describe a very mature radio system — DECT, or Digital Enhanced Cordless Telecommunication. Standardised in the mid-1990s, DECT now has a frequency allocation of 20 MHz around 1.9 GHz in almost every country (10 MHz in the USA). This allows up to 10 RF carriers, each bearing 1152 kbit/s using a simple GFSK modem. DECT then allocates 24 time-slots, each carrying 32 kbit/s of user data, plus various signalling and overhead data. In its original, cordless telephony application, one time-slot is used in each direction, delivering 3.4 kHz audio bandwidth with a simple ADPCM codec.
DECT telephony has proven highly successful, and DECT chips come pretty cheap at $2.50 or so in 10,000s. In addition, they are now very capable, integrating powerful processors with all the radio parts on a single device. Included on-chip is a capable DSP, which has allowed manufacturers to incorporate many additional functions such as telephone answering machines, modems for calling-line identification, echo cancellation for loudspeaker phones, better quality speech for VoIP and so on.
This confluence of advantages — dedicated spectrum, spectrum etiquette, robust interference mitigation, availability and low cost — makes DECT very attractive for wireless audio distribution. Indeed, the potential to develop a system that was far superior to existing solutions led Cambridge Consultants to look at developing a commercial solution. The result is a reference design, Salix.
The design process
Although DECT lends itself well to audio distribution, there were several factors that needed to be tailored in order to deliver the best possible performance. For example, for entertainment and all-day conferencing, the audio bandwidth must be better than 12 kHz. To provide this the open-source codec CELT was ported on to DECT chips from Dialog Semiconductor.
CELT, ‘Constrained Energy Lapped Transform,’ delivers very low latency whilst supporting stereo at 32 kHz sampling rate, with a fixed bit-rate of 64 kbit/s. To keep the bill of materials low, the CELT decoder is ported on to the on-chip DSP core in the receiver unit. A great deal of optimisation of the open-source code was needed to achieve this, but the cost saving per unit makes this worthwhile. The CELT encoder is more complex, and therefore runs on an external DSP in the transmitter board.
To deliver the required data rate the design uses a double time-slot in DECT, which allows for a 16-bit CRC word to protect each 64 bits of data. A packet loss concealment algorithm is triggered by the CRC pass/fail bits. Digital audio is then sent to an external DAC/headphone amplifier, which delivers stereo audio to a standard jack socket.
One of the ‘standard’ problems with a UHF radio is that of drop-outs in reception, as the receiver passes into a null resulting from multipath effects that occur in almost any environment. To overcome this issue the Salix receivers are designed with two antennas, and take advantage of DECT’s extended synchronisation sequence, which is long enough to allow each antenna to be tried in turn, and the one with the better signal selected. This space diversity improves link margins by about 10 dB, and is unusually cheaply implemented at a cost of one CMOS switch and less than 5 square centimetres of board.
One of the main advantages of DECT is of course that multiple systems can co-exist in one place due to a DECT’s robust spectrum etiquette scheme. In order to translate this capability to a wireless audio distribution application the new system uses the DECT functionality of first scanning all time-slots in all channels, building a map of received signal strengths, then choosing the quietest for use. This map is maintained in the background.
The receivers each send an occasional message upstream to report received signal quality. The system knows how many receivers are present (the transmitter broadcasts this number), and the receivers adjust their period between messages accordingly. A quality assessment algorithm in the transmitter uses the upstream messages to decide whether to move to another channel to improve performance. A seamless channel move is done by establishing the new channel first, then switching over, prior to clearing the old channel.
It was also important to ensure that the system is easy to manage, with straightforward connection. Connection of a receiver to a transmitter is done (for most conference and cinema use) by plugging the receiver into a configuration unit, and keying in the auditorium number, or other channel number, on this unit. A simple text-based serial interface into the receiver sets the identity of the transmitter to be used; the transmitter broadcasts this identity in a beacon. Alternative selection schemes, perhaps using the push buttons on the receiver, can easily be implemented.
Lastly, in addition to DECT allowing multiple co-located transmitters, the very simple GFSK radio modem means that a signal to interference ratio of only 9-10 dB is needed for an acceptable bit error rate. Hence DECT radio resource can be re-used after only a few tens of metres (or less in some types of building), allowing an essentially unlimited total number of systems in a large facility.
Tim Whittaker is System Architect at Cambridge Consultants – www.cambridgeconsultants.com.
