The Time Use Study we carried out in 2009/2010 showed that the current measuring system captures approximately 95% of today's radio listening and captures all consumption of terrestrial, cable, satellite broadcasting and DAB or DAB+. In other words, if someone listens to a conventional wireless or car radio, this will be registered. However, if someone listens to an Internet radio station or podcast after the programme in question is broadcast, this will elude the measuring device due to the requirement that programmes be broadcast and consumed at the same time. The reason for Audio Matching's inability to register the consumption of radio programmes streamed through the Internet lies in the dynamic transmission of the signal. In order to play a radio programme, a computer must first load the live stream, which produces a slight lag compared to on-air broadcast. Moreover, a computer, while playing a live stream, may sometimes have to buffer the signal, which causes a short interruption and an additional delay in transmission. Audio Matching is unable to process signals with varying delays.

Audio Matching as a means of station recognition requires not only simultaneity of the two original sound patterns but also an audible sound source. This means that the measuring device misses any listening that takes place with headphones.

We are working hard to remedy this. The data we are missing out on is of different magnitudes in different population groups, which increases the need for a solution. We are exploring several avenues. The two most promising ones are:
1. Expanding the measuring system
2. Record use in another way and add it up in the panel

1. Expanding the measuring system

As already mentioned, Audio Matching currently permits a signal delay of up to 3 seconds. However, as transmission via the Internet is delayed compared with normal radio, and this delay is not constant either, it is currently not possible to measure the use of Internet radio. One conceivable approach would be to increase the time window from 3 seconds to 60 seconds, which would thus also enable Internet streaming to be surveyed. However, this would entail a change in the attribution rules, as lengthening the recording increases the risk that canned music is erroneously recognized as radio. A further disadvantage of this approach is that it still does not enable the use of Podcasts and radio on-demand.

Another possibility is Audio Watermarking. This technology allows additional information to be embedded into a soundtrack without changing the way the track is experienced by listeners. Acoustic watermarks consist of sounds that, though audible, do not impede enjoyment. Radio stations and songs can be identified by the content of these watermarks. This makes it possible to register when anyone (re-)listens to radio programmes after they have been broadcast, e.g. when they listen to downloaded podcasts, paused DAB programmes, and radio on demand.
Audio Watermarking is not yet fully mature, as the audio watermarks are not absolutely resistant to compression, i.e. they may, under certain circumstances, be deleted when Podcast files are created. In addition, although the watermark is not audible to the human ear as such, the listener may in some cases perceive a “rattling” sound, as the watermark slightly changes the audio track. In addition, a full picture of the market can only be obtained provided that all channels, including all relevant foreign channels, participate.

Moreover, the concept of the "market" and the value thereof would have to be redefined: Where do we draw the line when reporting the "daily reach" of a programme, for example? Should we include someone who listens to a podcast before 6 am the day after a given programme is broadcast? What about someone who listens to the podcast a week later? And how do we differentiate between repeat listeners and first-time listeners? Should we review our core reach definitions?

 

2. Record use in another way and add it up in the panel

Another approach would be to record the non-collected usage data in another way and add it up in the panel. For instance, one possibility would be to collect data on Internet radio usage in collaboration with Net-Metrix and to then combine it with usage data from the radio panel.