Audio Matching for TV and Radio Audience Measurement
Publication Date: 2002-Aug-21
The IP.com Prior Art Database
1. SCOPE OF DOCUMENT
This document describes ideas for the development of an audio matching system for Television and Radio audience measurement applications. The techniques described how the system would be implemented; the ideas will be refined during further development and implementation.
2. THE PURPOSE OF THE SYSTEM
The system is intended to measure, as part of an otherwise-conventional TV audience measuring system, the TV or Radio audience for one or more of the following:
2.1 Broadcasts listened to at the time of the broadcast.
2.2 Broadcasts recorded and listened to later where the system can determine, at the time of playback, the time of the original broadcast. (This can be done, for example in the case of TV, by inserting a time and date code into the Vertical Blanking Interval of the video signal as it is recorded, and recovering said code on playback. This is a well-established technique.)
2.3 Broadcasts recorded and listened to later where the system cannot determine, at the time of playback, the time of the original broadcast.
2.4 Pre-recorded material played back in the home, such as bought or rented video cassettes or DVDs.
3. OVERVIEW OF THE METHOD OF OPERATION
The underlying system follows conventional TV audience measuring systems, with equipment in a panel of homes returning data about the use of TV and Radio equipment in those homes, typically by means of an overnight telephone call.
In a matching system such as this, audio samples are taken in the home representing what is being listened to by the panel members. The sounds may be picked up by one or more microphones, or by electrical connection to the TV or Radio.
Meanwhile back at base or, in the case of regional broadcasting at one or more remote locations, (hereafter termed reference sites) similar samples (hereafter termed reference data) are taken of all of the broadcast material of interest and, if relevant, of pre-recorded material of interest such as video tapes or DVDs.
Following retrieval of samples from each of the panel homes, typically overnight, the in-home samples are compared with the reference data in order to find a match, thus identifying what was listened to in each home.
The time at which the samples are taken is recorded, both in the home and at the reference sites. Thus for the situations described in 2.1 and 2.2 above the in-home samples may be compared directly with the relevant reference data, albeit with an allowance of a few seconds for inaccuracies inherent in the home and reference-site clocks. This is not the case for situation 2.3, where the whole bank of reference data, going back perhaps a week, may need to be searched before a match, if any, is found. Similarly for situation 2.4 the whole bank of reference data extracted from pre-recorded material may need to be searched.
4. PRACTICAL ISSUES TO BE ADDRESSED