2. Measurements of the sound levels of CMs as well as those of the programs

Transcription

1 Quantitative Evaluations of Sounds of TV Advertisements Relative to Those of the Adjacent Programs Eiichi Miyasaka 1, Yasuhiro Iwasaki 2 1. Introduction In Japan, the terrestrial analogue broadcasting is on the road to be replaced to the terrestrial digital broadcasting with a wider dynamic range than the analogue broadcasting by July in So, the programs are broadcast simultaneously in terrestrial digital broadcasting nearly identical to the conventional analogue programs in wider Japanese areas. As for conventional terrestrial analogue broadcasting, Kamada and Miyasaka [1], and, Miyasaka and Kamada [2] measured the sound levels of 2159 types of CMs recorded throughout a day for five Japanese TV terrestrial analogue channels. The results show that the averaged sound levels range from 2 db to 14dB, while the standard deviations of 45% of the CMs were less than 5 db. The maximum standard deviation was not more than 20dB. They also performed perceptual experiments in order to investigate the relationship between the physical sound levels and the corresponding auditory perception. They compared perceptually three different types of CMs such as CM 1(average : 5.1dB, standard deviation (SD) : 1.8dB), CM 2 ( 7.1dB, 4.9dB), CM 3 ( 14.2dB, 10.1dB) with the reference speech ( 6.5dB, 10.1dB). The results show that all CMs were perceived 3 db louder than the reference. This paper presents the results of measurements of the sound levels of 8846 advertisements (CMs) both in Japanese conventional terrestrial analogue broadcasting (TAB) and in the terrestrial digital broadcasting (TDB) for five main Commercial TV stations in Tokyo in order to compare the real dynamic ranges of CM sounds in TDB with those in TAB. 2. Measurements of the sound levels of CMs as well as those of the programs 2.1 TV channels to be investigated The channels to be investigated for the relative differences among daily averaged sound levels of Japanese TV channels were ten channels, shown in Table 1, consisting of five terrestrial analog broadcasting channels and five terrestrial digital broadcasting channels in commercial TV broadcasters. The sound levels of programs including TV commercials with the corresponding pictures for each TV channel were recorded with the 44.1 khz sampling and 16 bit quantization rate throughout a day for a week. The sound levels of actual main programs were calculated for 15 seconds just before the TV CMs, and also those of TV CMs with the durations of 15 seconds. 1 Professor, Faculty of Environmental and Information Studies, Musashi Institute of Technology 2 SEVEN & i HDGS. 3 69

2 Table 1 List of the investigated TV channels terrestrial analog broadcasting terrestrial digital broadcasting Nippon TV (ch 4) A N Nippon TV (ch 41) D N TBS (ch 6) A T TBS (ch 61) D T Fuji TV (ch 8) A F Fuji TV (ch 81) D F TV Asahi (ch 10) A A TV Asahi (ch51) D A TV Tokyo (ch 12) A K TV Tokyo (ch 71) D K 2.2 Number and durations of TV commercials investigated Table 2 shows the number and the durations of the TV CMs investigated. The total number of the TV CMs during the periods both for digital and analogue terrestrial broadcastings was x 2. The figure in parenthesis shows the figure of CMs obtained by excluding the same CMs. Table 2 he number and the durations of the TV CMs investigated TV channel Nippon TV TBS Fuji TV TV Asahi TV Tokyo Total Total (Analogue +Digital) Total Number of CMs Analogue/Digital (432) (474) (487) (410) (431) (2234) (4468) Total Duration Analogue/Digital 3:51:15 4:29:20 4:20:40 4:19:30 4:06:05 21 : 06 : : 13 : Sound levels of CMs Fig.1 and Fig.2 show the relationship between the averaged sound levels of CMs and the standard deviations in TAB and in TDB, respectively. The abscissa means the averaged sound levels of CMs, while the ordinate means the standard deviations. The total averaged sound level of the averaged sound levels of TAB and the standard deviation were 3.7 db and 5.9 db, respectively, while those of TDB were 5.4 db and 6.1 db, respectively. The averaged levels of CMs in TDB is about 2dB less than those in TAB, while the range of standard deviations in TDB is wider than in TAB. However, most CMs irrespective of TAB or TDB have relatively high averaged sound levels and small standard deviations. 70

3 Fig.1 Relationship between the averaged sound levels of CMs and the standard deviations in terrestrial analogue broadcasting (TAB) Fig.2 Relationship between the averaged sound levels of CMs and the standard deviations in terrestrial digital broadcasting (TDB) There are some differences between the TV stations. Fig.3 and Fig.4 show the results of TAB and of TDB at the Station A. There are few differences between them. Fig.3 Relationship between the averaged sound levels of CMs and the standard deviations in Terrestrial analogue broadcasting (TAB) at Station A 71

4 Fig.4 Relationship between the averaged sound levels of CMs and the standard deviations in Terrestrial digital broadcasting (TDB) at Station A Fig.5 Relationship between the averaged sound levels of CMs and the standard deviations in Terrestrial Analogue Broadcasting (TAB) at Station B Fig.6 Relationship between the averaged sound levels of CMs and the standard deviations in Terrestrial Digital Broadcasting (TDB) at Station B Fig.5 and Fig.6 show another results at the Station B. The averaged sound level of CMs in TDB is about 2dB less than that in TAB with wider standard deviation in TDB than in TAB. 72

5 3. PERCEPTUAL EXPERIMENT of CM SOUNDS Perception of CM sounds should not be solely investigated but be done relative to the adjacent main programs because the sound levels of the program segments prior to CMs may influence perception of the CM sounds. Kamada and Miyasaka [1] investigated perception of three different types of CMs with a reference speech uttered by a NHK female announcer using the transformed up down method. The results show that the CMs used in the experiment were perceived about 3dB louder than the reference (the averaged sound level : 6.5dB) although they had different averaged sound levels such as 5.1 db, 7.1 db, and 14.2dB, respectively. The authors performed an additional experiment of perception of CM sounds relative to the real sounds of the adjacent programs. 3.1 The stimuli 128 CMs among all CMs were selected which were perceived noisy or louder compared to the adjacent programs by the authors. From the selected CMs, three CMs with the duration of 15s for each TV station were used as the stimuli. The total number of the stimuli was 15 (= 3x5stations). The left columns of Table 3 show the averaged sound levels of the programs for 5 seconds just before the first CM is inserted and the standard deviations, while the center columns show those of the first CM for 5 seconds just after it is inserted and the standard deviations. The right columns show the sound level differences between the programs and the first CMs. Stimulus Number Table 3 Characteristics of the stimuli used in the perceptual experiment Averaged sound levels : A [db] Programs Standard Deviations [db] Averaged sound levels : B [db] CMs Standard Deviations [db] Difference B A[dB] Fig.7 shows an example of a power envelope consisting of program segments with the duration of 73

6 15s and the subsequent CM segments. A silent period can be found between the program and the CM. Fig.7 An example of power envelope consisting of program segments with the duration of 15s and the subsequent CM segments. 3.2 Experimental Procedure After the observers were indicated to adjust the sound level of each program to a comfortable level before evaluation, they were required to evaluate each CM sound with 5 point scale from quiet to noisy, comparing with each program sound. 3.3 Results of the experiment Fig. 8 shows the rate of the answers for the stimuli. The black bar indicates the percentages of the observers who perceived the corresponding CM sound as very noisy. The gray bar shows the percentages for the rank order 4, the white bar for the rank 3, the dotted bar for the rank 2, and the shaded bar for the rank 1 quiet. For example, for the stimulus no.13, 67% of the observers perceived very noisy, 25% of them noisy and so on. Fig.8 The rate of the answers for the stimuli. 74

7 Fig.9 The differences of sound levels between the adjacent programs to CMs with the duration of 5s Fig.9 shows the differences of the averaged sound levels between the programs and the sequent CM with the duration of 5s. Black bars indicate the differences of sound levels. For the stimulus No.8, the total percentage for the rank 4 and 5 is over 95% which indicates noisy although the sound level of the CM during the first 5s is only 2 db more than that of the program. Similar results can be found for the stimulus no.2 in which the total percentage for the rank4 and 5 is over 70%, while the CM sound level is smaller than the program level. These results infer that the averaged sound levels can not always reflect correctly the auditory impression. Recently, modified B weighted LAeq measure of sound level has been presented to perform well in predicting auditory impression [2]. The results of our experiments presented in this paper, however, imply that an additional modification of measure of sound level may be necessary. 4. Discussion and conclusion First, qualitative evaluation was conducted for advertisements (CMs) in Japanese conventional terrestrial analogue broadcasting (TAB) and in the terrestrial digital broadcasting (TDB). The averaged levels of CMs and the standard deviations in TDB are about 2dB less than those in TAB. Although TDB has a wider dynamic rage than TAB, most CMs have high averaged sound levels and small standard deviations irrespective of TAB or TDB. Secondly, the perceptual experiment was performed in order to investigate the relationship between the physical sound levels and the perceived auditory impression. Using 15 CMs selected through a preliminary perceptual experiment by authors, observers were required to judge each stimulus with 5 point scale. The results imply that the averaged sound levels do not always correspond to auditory impression. There will be another serious problem on TV CMs as well as the sound levels. A way of insertion of CMs into main programs is one of the problems. For example, it is not unusual that advertisements (CMs) in Japanese broadcasting are abruptly inserted into a main program without any message, even though a program anchorperson sometimes announces now for CMs in news programs. Abrupt insertion sometimes interrupts concentration of viewers on TV programs such as drams, documentaries, serious news programs and so on. Miyasaka and Kimura have started the 75

8 ways of insertion of TV CMs into main programs comparing with foreign countries [4]. Acknowledgement A part of this research was supported by SCAT (Support Center for Advanced Telecommunications Technology Research) (2005, 2006, 2007). References [1] Takahiro Kamada and Eiichi Miyasaka, Relationship between the sound levels of TV CMs and Auditory impression, Proceedings of Spring meeting of Acoust. Soc. Japan (Japanese), 1 3 2, (2006) [2] Miyasaka and Kamada, Effects of TV commercials inserted in main programs on Viewers, Bulletin of Faculty of Environmental and Information Studies, Musashi Institute of Technology, Vol.8, pp (2006) [3] AES Staff Writer Loudness Trumps Everything, J.Audio Eng.Soc.,Vol. 54, No. 5, pp , 2006 [4] Miyasaka and Kimura, Sound levels of TV advertisements relative to the adjacent programs and cross national comparison of the way of their insertion into programs, AES 122nd convention, paper 7045, (2007, Vienna) 76