Audio Loudness Normalization and Measurement of Broadcast Programs Martin Simaliak, Miroslav Uhrina, Jan Hlubik, Martin Vaculik, Martin Cap Department of Telecommunication and Multimedia Faculty of Electrical Engineering, University of Zilina Zilina, Slovakia martin.simaliak@fel.uniza.sk, miroslav.uhrina@fel.uniza.sk, jan.hlubik@fel.uniza.sk, martin.vaculik@fel.uniza.sk Abstract This paper specifies the method of the subjective evaluation and present the result of measurements. The measurements mapped actual state of the current media sector in the region of Slovakia. In today`s media landscape occurs despite the standards complied with recommendations on the issue of normalization volume. The measure data from a sample of various program channels are evaluated and appropriately considered. The evaluation follows the standard EBU recommendation 128. The results are presented in term of preferred gating parameters. The results are based on proposal for technological solutions that lead to improved standards compliance of loudness normalization. Older standard BS.1770 defines the measurement method and latter recommendation R128 defining a specific target level for loudness normalization. Keywords- EBU R128; loudness normalisation; peak level; loudness range I. INTRODUCTION Undesirable loudness jumps are a notorious problem in television or radio broadcast. In this paper we will discuss the EBU recommendation R128, a new and open standard for balancing audio programs according to their actually perceived loudness. We are dedicated to problems such as jump in audio levels at the breaks within programs, between programs and between channels. Loudness perception is complicated. The physical sound pressure level tells very little about sound loudness which will be perceived by a listener. The solution consist in switching to loudness based metering and program normalization. Unbalanced audio loudness falls into a specific area designated as the loudness war. The lack of proper loudness control has enabled this problem between commercials, which have consequently become hyper compressed in order to be as loud as possible thus sacrificing good sound quality. Based on the research needs of the production, distribution and preparation for broadcasting audio signals, and existing methods of measuring and monitoring the EBU define the causes of the problem of uneven loudness, and based that issued recommendations for normalization and loudness maximum permissible level of audio signals for broadcasting. At present, these standards for loudness normalization are existing: Standard defined by European Broadcast Union (EBU) defines in the R128 document consist from these technical data sheet [1]: EBU Tech Doc 3341 [2] EBU Tech Doc 3342 [3] EBU Tech Doc 3343[4] EBU Tech Doc 3344[5] Standards defined by International Telecommunication Union (ITU) for loudness normalisation: ITU-R. BS.1770, BS.1770-1, BS.1770-2 [6] II. PROBLEMS CAUSED BY UNEVEN LOUDNESS Analysis of existing methods for measuring and monitoring audio broadcasting, as well as the needs of modern digital production can be defined by the following cause s problem of uneven loudness: Normalization of the peak audio (normalization to the peak time) led to a significant difference in loudness between different program content and between different broadcasters. Inconsistency between loudness and programs channels is the cause of most complaints listeners. Monitoring of the peaks in the usual way, use QPPM (Quasi-Peak Programme Meter) instrument specified by EBU Tech Doc 3205-E [7] cannot monitor the loudness, as QPPM not designed to indicate a longtime RMS signal. With the development of digital production, distribution and transmission system, the maximum allowed level of audio signal specified in ITU-R BS. 645 is more adequate. Based on the above troubleshooting uneven loudness is based on the change of the concept of measurement and monitoring the audio signal that is the subject of the documents EBU [7]. III. BASIC PRINCIPLES OF LOUDNESS NORMALISATION The analysis of the causes of the problem of uneven loudness has basic principle of loudness normalization: Audio broadcasting is normalized to loudness, and not in relation to the peak value, and measurement and monitoring the audio signal can not be done QPPM instruments, but instrument that measure the loudness (Loudness meter) and the actual peak value (True Peak Level meters). - 531 -
Recommendation EBU R128 loudness descriptors defined audio signal that can be measured and monitored, and recommended values of these descriptors with the loudness normalization. According to EBU R128 can define the following loudness descriptors and their recommended values: Programme Loudness(Integrated loudness, IIk) Dynamic range of loudness (Loudness Range, LRA) The maximum true peak level (True peak, TP) The Programme Loudness Level shall be normalized to -23 dbfs (Loudness Units in respect Full Scale). Deviations from this value, it should to be greater than ± 1 LU (Loudness Unit) programs where not possible to achieve the program loudness is true -23 LUFS (e.g. programming material which is live broadcasting). The dynamic range of loudness is a parameter calculated according to the algorithm defined in [3]. The maximum permissible level of actual peak values (True Peak) program material is -1dBTP (db True Peak) and measures the instrument described in ITU-R BS. 1770 and EBU Tech Doc 3341. In addition to the loudness descriptors, EBU documents defined and following postulates: The audio signal should be observed, measured and monitored as a whole, rather than as a collection of different elements as signal as speech, music or sound effect signal. Measurement and monitoring of audio signals to perform loudness compatible with the ITU-R 1770 and EBU Tech Doc 3341. Audio systems, signal processing, and all operations related to the production of program material should comply with EBU Tech Doc 3343 Audio systems, processing and all operations of the program material relating to the distribution of the signal should be comply with EBU Tech Doc 3344. The algorithm calculating of loudness is defined by International Standard ITU-R BS.1770, and the unit loudness of the LU or LUFS [7]. IV. EXPERIMENT In order to assess the situation in Slovakia, the measurement was done three TV stations and three radio stations that are in territory of Slovakia. The measurement was conducted using the software meets the requirements EBU. For the purpose of measuring was used the following equipment: TV with satellite receiver and Radio Computer with software for loudness monitoring Sound card for computer How the receiving device was used a satellite TV and radio. The goal of this analysis is the measurement of loudness differences between different TV and radio stations and between contributions in real conditions. Measurement procedure: Deactivating all automatic loudness adjustment on TV and radio, which corrects the loudness of contribution to one level. Set input bus on the level at which the signal peaks reached the maximum accessible level, which is for TPM -6 dbfs Recording values Size entry level remained the same and record the data on the loudness level for individual contribution as: TV Production, TV News, TV Serial and TV Advertising Input level was the same for measuring individual contribution to various TV station Method of recording for radio broadcast was the same as for measuring the loudness of TV broadcasting has been set desired maximum peak level. The values of the loudness level of individual contributions such as broadcast from the studio, report, transportation service, music section and advertising was recorded for more radio stations in the preset one input loudness level All measurement were performed several timed and the results are based on averages of individual measurements. To evaluate the loudness of the contributions, we used data from a record, such as a maximum peak level for the left and right channels, the average loudness and the LRA. The level peak is shown in relative scale for TPM meter, the maximum level is 0 db corresponding to the absolute level of -6 dbfs. The volume level LK (K- Weighting) is also shown in relative scale, the relative level is 0 LU, the absolute level 0 LU= -23dBFS. Loudness range LRA is in units of LU. Readings were processed and evaluated and are listed in the following tables and graphs. TABLE I. NUMBER 1. - 532 -
TABLE III. NUMBER 3. Figure 1. Compare the loudness of contributions in the same loudness level and the alignment at the same volume level for the TV station number 1. TABLE II. NUMBER 2. Figure 3. Compare the loudness of contributions in the same loudness level and the alignment at the same volume level for the TV station number 3. TABLE IV. NUMBER 1. Figure 2. Compare the loudness of contributions in the same loudness level and the alignment at the same volume level for the TV station number 2. - 533 -
TABLE VI. NUMBER 3. Figure 4. Compare the loudness of contributions in the same loudness level and the alignment at the same volume level for the radio station number 1. TABLE V. NUMBER 2. Figure 6. Compare the loudness of contributions in the same loudness level and the alignment at the same volume level for the radio station number 3. TABLE VII. THE VALUE OF LOUDNESS LEVEL FOR DIFFERENT GENRES. Figure 5. Compare the loudness of contributions in the same loudness level and the alignment at the same volume level for the radio station number 2. - 534 -
the maximum permissible level. When setting the loudness level to the desired reference level -23 dbfs, the peak level for all radio stations was in accordance with the maximum allowable level. Figure 7. Compare peak level and loudness level with the loudness range for different genres In TV broadcasting mean loudness of individual contributions from one TV station differently, loudness level were different even between different TV stations, which is not in accordance with the loudness normalization to the desired reference level. Most excited posts in all measured stations were advertising. The difference between the loudness of contributions result in the permissible excellence in individual contributions. The graphs show the top level with an average loudness. The graphs show the size of the excellence of the individual contributions at different volume levels. The chart also show how level of excellence achieved individual contributions at the same volume level. Contribution may not exceed the permissible level of excellence, but when was set the loudness to the reference level of the peak level was greater than s is permitted. This may be the opposite, that the contribution exceeded the permissible level of excellence, but when you set the loudness to the reference level of excellence is less than desired maximum level. Loudness levels of contribution shown in the graphs, which are aligned to the desired level of -23 dbfs in some cases exceed the permissible level of excellence for example in TV station 1. This may also be due to the fact that the TV station has selected reference level of -24 dbfs. Reference level, according to sources can be defined in the range -22 to -24 dbfs. We choose level -23 dbfs. Limitation of peak in this case cannot be considered, because broadcast contribution wasn t normalization to a desired reference level adio station has small differences in the average loudness of contribution. The smallest fluctuations in loudness amounted radio station number 3, but the peak level exceeded V. CONCLUSION The measured loudness levels of the various contributions to the various TV and radio stations are shown in tables and graphs. The measured loudness levels of the various contributions to the TV station were differemnt, which is not in accordance with the desired loudness normalization to the reference level. In contrast, the radio station had a much smaller difference in the loudness contributions.the maximum permissible peak level was exceeded several times during TV broadcasts, but also adjusted to the reference loudness level, top level exceed the maximum allowed. In radio broadcasting peak level exceeded the allowed maximum level, but after adjusted to the reference loudness level peak level was in accordance with the requirements of the maximum permisible peak level. Measuring the loudness level, loudness range and peak level was aslo performed for each genre, in which were verified knowledge about the difference in the ratio of the peak level to loudness level and differences in the loudness range. The analysis of standards for the normalization volume was determined by reference loudness level -23 dbfs. According to this reference level were processed measured loudness level. ACKNOWLEDGMENT This work has been supported by: Centre of excellence for systems and services of intelligent transport, ITMS 26220120050 supported by the Research & Development Programme founded by the ERDF. REFERENCES [1] EBU R 128, European Broadcast Union, Geneva, 2011. [2] EBU Tech Doc 3341, European Broadcast Union, Geneva, 2011. [3] EBU Tech Doc 3342, European Broadcast Union, Geneva, 2011. [4] EBU Tech Doc 3343, European Broadcast Union, Geneva, 2011. [5] EBU Tech Doc 3344, European Broadcast Union, Geneva, 2011. [6] ITU-R 1770-2, International Telecommunication Union, Geneva, 2011. [7] M. Mihajlovic, D. Todorovic, Normalizacija glasnosti programskog materijala, August 2011, Belgrade, pp. 1111-1114, ISBN: 978-1-4577-1499-3. - 535 -