ITU-R REPORT BS.1204

Abstract

<u>Introduction</u> <br>Whenever the video and audio components of a television signal are transmitted over separate paths, there is the possibility that their transmission times will be different. When the difference is such that the viewer perceives a separation between vision and sound, enjoyment of the programme can be impaired. The subject has been studied in detail in Australia and is reported in [CCIR, 1986-90a]. <br>A guide to the amount of allowable time difference, may be taken from the standards for film, where the picture can lead the sound by no more than two frames (83 ms) or lag behind it by no more than one frame (42 ms). The reason for the asymmetry in these allowable time differences is that sound arriving after vision is a familiar experience while sound arriving before vision is not. <br>Frame synchronizers are used increasingly in long distance transmissions because they confer so much operational flexibility. However, large delay differences can easily be introduced unless steps are taken to have the sound delayed equally with the vision within the synchronizer. <br>Delay differences are often compensated manually by an operator trying to judge the correction before other viewers can perceive it, but this is a difficult process. A method is under study by which the relative delays between vision and sound can be corrected automatically. <br>An audio signal in digital form is easily delayed be feeding it into a shift register and the delay can be varied by changing the address from which the output is taken. The scheme shown in Figure 1, only provides an increased delay of the audio component. <br>A similar delay of video is more difficult to achieve. However there is little need for such a video delay because sound following vision is more easily tolerated, and is less likely to occur. The only foreseeable cause of increased delay in the audio path is the insertion of a digital FIR low-pass filter with a sharp cut-off. Such a filter might be needed, for example, in converting the audio sampling rate from 48 KHz in the studio to 32 KHz for transmission, but this would only introduce a delay of 5.3 ms.