On Levelling and Loudness Problems at Television and Radio Broadcast Studios

The problem of partially extreme loudness differences in radio and television programmes has been well known for a long time. With respect to the introduction of new digital techniques combined with parallel transmission of digital and analogue signals the problem of loudness differences again is especially significant. Based upon relevant levelling recommendations and a newly developed loudness algorithm solutions avoiding loudness differences in radio and television are presented. 1. PREFACE Loudness leaps in radio and television sound increasingly bother listeners. Those loudness leaps are extremely obvious when zapping through radio and television programmes of European digital DVB channels. Within one programme, the transition from a film dialog to a strongly compressed commercial is perceived as particularly bothering. Both underlevelling and over-levelling are observed, resulting in level differences of more than 15 dB. Among others, the reasons for such levelling and loudness leaps are as follows: • Apparent obscuration concerning levelling of sound channels • Usage of particular different and partly unspecified programme level meters • No standardized studio loudness meter is available until now • archive material (analogue and digital) is partly not adapted to the regarding sound channel In most of the analogue radio FM-channels the loudness is mainly balanced by means of compressors and limiters meeting the permissible maximum frequency deviation. This situation seems to be tolerated momentarily. Considering digital broadcasting it should be possible to achieve balanced loudness profiles when regarding KLAR AND SPIKOFSKI ON AUDIO LEVELLING AND LOUDNESS PROBLEMS AES 112 CONVENTION, MUNICH, GERMANY, 2002 MAY 10–13 2 the existing international recommendations of ITU and EBU. This target should be aspired regarding the comparison of different programmes as well as different contributions within one programme. 2. CHARACTERISTICS OF AUDIO PROGRAMME METERS 2.1 ALIGNMENT LEVEL ITU Recommendation ITU-R BS.645-2 [3] defines the levelling of radio channels by means of the “Alignment signal (1 kHz sinus)”. The actual specified level of the sinus signal refers approximately to a full scale programme level regarding loudness. As the alignment signal has to be considered as a “static” signal, it can be measured by means of usual RMS-meters as well as specified programme meters. It has to be considered that the analogue “alignment level” and respectively the “nominal” or “permitted maximum level (PML)” is specified diversely considering national and international recommendations (Tab. 1). Recommendations for analogue & digital Audio Levels Alignment Level AL -9 dBr (35%) Nominal Level PML*** 0 dBr (100%) ITU-R BS.645-2 Transmission Level international 0 dBu* +9 dBu ARD HFBL-K Studio Level national -3 dBu (adaption) +6 dBu (adaption) US (UK) Reference Level national +4 dBm (dBu) (adaption) EBU international digital Transmission & Studio Level 18 dBFS 9 dBFS ***) PML = Permitted Maximum Level *) 0 dBu = 0.775 V rms (sine) = 1.1 V peak 0 dBFS = Clipping Level (FS = Full Scale) Tab. 1: Audio levels in transmission and studio environments Regarding digital audio channels, the relation between the alignment signal and the full scale or clipping level was already specified in 1992 EBURec. R68 [8]. (Tab. 1). Following this recommendation the difference between full scale or clipping level and alignment level has to amount to 18 dB. In other words the alignment level has to be – 18 dBFS. 2.2 AUDIO PROGRAMME METERS FOR BROADCASTING Today a lot of different programme meters are in use at professional studios with widely varying ballistically features. Whereas in America and Australia mainly VU-meters [5] are used, for European countries “peak programme meters (PPM)” are recommended by the EBU [10]. Those PPM are specified in corresponding IEC-recommendations IEC 268-10 [4] (analogue PPM) and IEC 268-18 Digital PPM [6]. The IEC-category of PPM are so called “Quasi-peak programme meter QPPM” which neglect short signal attacks due to the human ear. For digital PPM EBU recommends the same ballistcs as described in IEC 268-10 (Type 1). Since the introduction of digital audio techniques in broadcast, additional but not precisely specified PPM’s have caused some confusion. The following table shows those PPM’s which are in use in Europe at the moment. Besides their different scale-layouts, the PPM primarily vary with respect to their ballistic features described by parameters like “attack time” or “integration time” and “fall back time” or “decay time” (Tab. 2). Considering the scale-layout the full scale-tag (100%-tag = 0 dB) as well as the specified headroom should accord to “the attack time” of the regarding programme meter. As an example, the VUmeter, which can be considered as relatively slow, obviously needs an appropriate headroom because of the invisible signal peaks. Consequently the difference between the 100%-tag and alignment level has to be smaller than in other cases. Therefore recommendation ITU-R BS.645-2 [3] specifies the 100%-tag to be equal to the alignment level. In contrast to VU-meter the 100%tag amounts to a 9 dB higher level regarding QPPM. Remark: The attack time of the PPM, which is used in German broadcast (ARD and ZDF) [15], is KLAR AND SPIKOFSKI ON AUDIO LEVELLING AND LOUDNESS PROBLEMS AES 112 CONVENTION, MUNICH, GERMANY, 2002 MAY 10–13 3 specified as 10 ms / 90%. That means it takes 10 ms to reach the 90%-tag. The IEC-type, which is used by the BBC, is slightly different specified as 10 ms / 80%. Regarding fast digital “sample programme meter SPPM” theoretically no headroom is needed. Those meters are appropriate to control signal peaks with respect to clipping but they are not as suitable as QPPM regarding adequate programme levelling. For example, signals with high proportion of peaks tend to be under-levelled whereas strongly compressed with limited peaks tend to be over-levelled. This can result in grave loudness leaps, which seem to be more intensive than using a QPPM. The use of unspecified level meters is widely observed in digital audio fields. Mainly if sound engineers are familiar with specified level meters, the use of unspecified devices could result in severe levelling mistakes like clipping and loudness leaps. Because of the wide spread of characteristics of unspecified instruments it is difficult to get familiar and gather experience in levelling. Digital programme meters frequently are software applications. As known from those applications there are ”infinite” error sources. Whereas the “attack time” normally is almost 0 ms, that means the peak samples are indicated correctly, but there is a wide variation concerning the decay time. Those effects can result in different display as well as in different levelling. In Germany the QPPM is precisely specified in ARD-Pflichtenheft 3/6 [15]. The specified meter is recommended for levelling analogue and digital signals. Three additional PPMs with 0 ms, 0.1 ms and 1 ms attack time, which are also specified here, should only be used for controlling and not for levelling. In order to avoid confusion the IRT suggests to adapt the scale layout of digital PPM to the scale layout of the analogue QPPM [4] (Tab. 2). That means the 100%-tag is e. g. 10 dB below full scale. 3. DYNAMIC RANGE OF DIGITAL AUDIO SYSTEMS 3.1 PROGRAMME LEVELLING AND HEADROOM As already mentioned, the levelling range and necessary headroom depends on the ballistical features of the meter in use. Whereas VU-meters need up to 18 dB headroom, corresponding PPMs only require 9 – 10 dB [3]. The EBU-headroom of 9 dB is strictly combined with QPPM according to [3] and it’s reference to the alignment level specified in [8]. Using instruments with different ballistical features obviously result in other headroom recommendations. Headroom has to be considered as a buffer range between nominal and clipping level. Meeting the European recommendation, the exchange of programme material is guaranteed without any levelling problems. German broadcasters agreed to this recommendation and specified the headroom in ARD HFBL-K Rec. 15 IRT [16] according to the EBU-recommendation. Regarding analogue signals and devices including A/D and D/A-conversion, the absolute level limit at German broadcast studios is +15 dBu (100%-tag = +6 dBu + 9 dB headroom) (Fig. 2). 3.2 USABLE DYNAMIC RANGE OBJECTIVE AND SUBJECTIVE CONSIDERATION Discussing the specifications of headroom and footroom naturally the question whether the resulting system dynamic is sufficient with respect to the human ear has to be answered. In other words, which quantization or how many bits are necessary to guarantee a transmission of music signals without perceivable noise. To answer this question it can be referred to a responding paper from 1985 [25]. In the following the conditions and results of this study are shortly presented. Audiobitrate reduction systems, as for example Minidisc (ATRAC), MPEG 1, Layer 2 (MP2) and Layer 3 (MP3), not being introduced during the above mentioned investigation, are not considered in this context. Compared to linear PCM-systems (Pulse Code Modulation) these systems need evidently less quantization. The fact that they nevertheless allow noise-free recordings show that in the cases of bitrate reduction systems other quality parameters have to be considered. Regarding PCM-systems, the system dynamic is defined as the level differences between full scale programme level and system inherent noise level. The system dynamic, the signal-to-noise ratio or quantization noise can be calculated by means of the following formula S / N [dB] = 6n + 2 (n = quantization / bit). KLAR AND SPIKOFSKI ON AUDIO LEVELLING AND LOUDNESS PROBLEMS AES 112 CONVENTION, MUNICH, GERMANY, 2002 MAY 10–13 4 The calculated value – with negative sign – corresponds to the RMS-value of the quatization noise related to 0 dBFS programme level