Television audience measurement systems come in several flavors, but the basic structure involves a data signal that captures and analyzes what programs people are watching. Originally, data was inserted into the vertical blanking interval of the analog video signal, but in digital television, the VBI is not preserved, so an alternate method must be employed.
Ratings encoders for DTV use audio watermarking to insert data into the program audio stream in a way that very few, if any, viewers can detect. Based on careful psychoacoustic modeling, data is carried during scenes that will mask its presence from audibility, but not from being detected by special measuring gear placed in target viewers’ homes. There are two ways this is done today. The first is by inserting a watermark into the baseband audio signal; the second is to insert the watermark into a precompressed audio bitstream.
To minimize any potential damage to the ratings information, it is important that the transmission audio gear be arranged so this data is preserved on its way to the consumer. In the baseband approach, experience has shown that the ratings encoder should be installed just before the Dolby Digital (AC-3) encoder, after any other processing gear. In the case of the pre-compressed approach, the only place the ratings encoder will fit is after the Dolby Digital encoder, and usually after the DTV multiplexer in the transport stream. This method is very robust and any upstream processes are very unlikely to cause any problems.
