1、 CEA Bulletin NTSC/ATSC LOUDNESS MATCHING CEA-CEB11-A September 2009 Copyright Consumer Electronics Association Provided by IHS under license with CEA Not for ResaleNo reproduction or networking permitted without license from IHS-,-,-NOTICE Consumer Electronics Association (CEA) Standards, Bulletins
2、 and other technical publications are designed to serve the public interest through eliminating misunderstandings between manufacturers and purchasers, facilitating interchangeability and improvement of products, and assisting the purchaser in selecting and obtaining with minimum delay the proper pr
3、oduct for his particular need. Existence of such Standards, Bulletins and other technical publications shall not in any respect preclude any member or nonmember of CEA from manufacturing or selling products not conforming to such Standards, Bulletins or other technical publications, nor shall the ex
4、istence of such Standards, Bulletins and other technical publications preclude their voluntary use by those other than CEA members, whether the standard is to be used either domestically or internationally. Standards, Bulletins and other technical publications are adopted by CEA in accordance with t
5、he American National Standards Institute (ANSI) patent policy. By such action, CEA does not assume any liability to any patent owner, nor does it assume any obligation whatever to parties adopting the Standard, Bulletin or other technical publication. This CEA Standard is considered to have Internat
6、ional Standardization implication, but the International Electrotechnical Commission activity has not progressed to the point where a valid comparison between the CEA Standard and the IEC document can be made. This Standard does not purport to address all safety problems associated with its use or a
7、ll applicable regulatory requirements. It is the responsibility of the user of this Standard to establish appropriate safety and health practices and to determine the applicability of regulatory limitations before its use. (Formulated under the cognizance of the CEAs R4 Video Systems Committee.) Pub
8、lished by CONSUMER ELECTRONICS ASSOCIATION 2009 Technology Phone 202-872-9160; Fax 202-872-9161; Internet http:/www.atsc.org/stan Phone 800-854-7179; Fax 303-397-2740; Internet http:/; Email IEC Central Office, 3, rue de Varembe, PO Box 131, CH-1211 Geneva 20, Switzerland; Phone +41 22 919 02 11; F
9、ax +41 22 919 03 00; Internet http:/www.iec.ch; Email pubinforiec.ch FCC Bulletins: Federal Communications Commission,445 12th Street, SW,Washington, DC 20554 ITU Documents: ITU documents are available through the web at http:/www.itu.int/opb/ 3 Glossary ADC Analog to digital converter. compr A meta
10、data control word contained in AC-3 bitstreams that introduces limiting when the decoder is in the RF operating mode, and downmix overload protection in all modes. DAC Digital to analog converter. dBFM dBFS Acronym for decibels with respect to full scale. 3Copyright Consumer Electronics Association
11、Provided by IHS under license with CEA Not for ResaleNo reproduction or networking permitted without license from IHS-,-,-CEA-CEB-11-A dialnorm A metadata control word contained in AC-3 bitstreams that adjusts or normalizes decoder gain to achieve normalized loudness or “dialog” level. Leq(A) A meth
12、od of measuring audio loudness using an A weighted integrating-averaging sound level meter. LKFS Method for measuring audio loudness using a frequency-weighted integrating-averaging measurement as per ITU-R BS.1770.“ PCM Abbreviation for Pulse Code Modulation. PCM is sometimes used in the industry t
13、o describe baseband digital audio signals or the audio output format. RMS Root mean square. MPEG Used in this document to refer to the MPEG 1 Layer II digital audio rate reduction codec, defined by the ISO/IEC 11172-3 Ed. 1.0 standard 4 Background Consumer broadcast products have evolved to a point
14、where multiple services may be received by a single set top box (STB) or integrated receiver. Home users tend to set speech levels to a preferred loudness level. Speech is an essential component of virtually all broadcasting. Therefore it is desirable for different program sources (NTSC, ATSC, MPEG
15、etc.) to deliver an approximately constant speech level so that the listener does not need to adjust the volume control for different sources. Loudness measurements may be made by either of two methods. The Leq(A) method preceded the LKFS measurement which was developed to estimate the subjective lo
16、udness of a wider range of program material than that addressed by the Leq(A) method. For the purposes of this document, either method is acceptable, as they both give a good estimate the loudness of speech. The AC-3 (also known as Dolby Digital) audio system carries a metadata control word, dialnor
17、m, that varies the gain in the decoder and that is set during encoding so that speech is reproduced at a normalized level from the audio outputs of the decoding device. In Line operating mode, that level is 31 dBFS Leq(A) or LKFS in one channel (e.g. center front of a 5.1-channel source) or equivale
18、ntly for two-channel stereo, at 34 dBFS in the Left channel and -34dBFS in the Right channel. In RF operating mode, it is 11 dB higher, 23 dBFS for both channels. The gain structure described herein is based upon the condition that the AC-3 audio program is encoded with the dialnorm value set to cor
19、rectly correspond with the measured dialog level of the program material. The gain structure goals are also only achieved if NTSC (analog terrestrial) and MPEG-1 Layer 2 (digital direct to home satellite) audio program levels were set in accordance with currently accepted broadcast practices for Nor
20、th America. 5 Inputs 5.1 NTSC Current NTSC operating practices “process” audio by compressing and limiting the audio to control peaks and to reduce the peak to average level ratio. This allows peaks to consistently reach, but not exceed, the maximum permitted level, and has the side effect of normal
21、izing the loudness of speech in television programming. Measurement of a large number of channels indicates that program peak levels are typically 2 dB below 100% modulation after de-emphasis. The subjective loudness is typically 17 dB Leq(A) or LKFS below 100% modulation either in mono or with both
22、 stereo channels summed to mono. NTSC broadcasting has strict limits on maximum deviation of the aural carrier. These limits are not significant for RF re-modulators in set-top boxes and VCRs. Typical NTSC television tuners receiving the output of a mono re-modulator are not expected to overload unt
23、il the deviation is more than 8 dB above the nominal FCC mono (baseband) limit of 25 kHz. On the other hand, the deviation of a BTSC (stereo) re-modulator extends to 50 kHz. The pilot carrier adds another 5 kHz of deviation, leaving only a few dB to accommodate the occasional uncontrolled excursion.
24、 4Copyright Consumer Electronics Association Provided by IHS under license with CEA Not for ResaleNo reproduction or networking permitted without license from IHS-,-,-CEA-CEB11-A The BTSC demodulator output level should be set to 500 mV per the EIAJ CP 1203 standard (-12 dB with respect to the 2 Vol
25、t maximum output voltage) when receiving a 400 Hz L+R signal modulated to 12.5 kHz p-p subcarrier deviation. The 5 dB attenuation shown in Figures 1, 2 and 3 brings this level to -17 dB with respect to 2 Volts. 5.2 AC-3 in ATSC Digital Television An AC-3 decoder has two operating modes, Line and RF.
26、 Line mode provides the original dynamic range (and offers the listener various dynamic range control options). The dynamic range of the Line mode is potentially much greater than that of NTSC. If Line mode is used for driving an RF re-modulator, average audio levels will be lower to allow for the u
27、nrestricted dynamic range. The RF mode of the AC-3 decoder, however, is specifically designed to drive RF re-modulators. It offers a restricted dynamic range, and boosts the level of the program by 11 dB. This combination allows ATSC broadcasts to sound similar to NTSC broadcasts. AC-3 has greater h
28、eadroom above speech peaks than NTSC, potentially 6 dB more in RF mode1and 17 dB more in Line mode. The latter allows the transmission of virtually any source (including wide-range movies) without the need for compression or limiting and with the speech normalized; depending on the equipment, the li
29、stener may have choices in the actual reproduced dynamic range. In RF mode, the metadata word, compr, introduces limiting when necessary to prevent clipping. In either operating mode, it is possible for peaks to reach full-scale. It is important to note the AC-3 decoder does protect for overload dur
30、ing RF modulation when the signal is down mixed to stereo or mono in a set top box or other consumer decoder set to line level. In Line mode, program peaks can be 11dB larger than the average loudness level. To avoid extreme overload within the modulator or tuner, the program average level must be 1
31、1dB lower that is for RF Mode levels. This results in the AC-3 decoder normalized dialog level of -34dBFS in each of two stereo channels, which is equivalent to -28dBFM. The analog line output shown in Figures 1, 2 and 3 should produce 2 Volts RMS when driven by a 0 dBFS signal. The variable attenua
32、tor shown preceding the analog / digital selector switch passes the AC-3 Line mode sourced signal through. It attenuates the AC-3 RF mode sourced signal by 11 dB to eliminate the boost applied by the AC-3 decoder in this mode, leaving the average program (or speech) level at -34 dBFS. The simplest w
33、ay to drive a Mono re-modulator is to set the modulator sensitivity so that a 0 dBFS signal will produce 50 kHz deviation (200% modulation). An AC-3, RF mode sourced signal will then match the loudness of an NTSC signal, while a Line mode signal will be lower in average loudness, as required to hand
34、le the original dynamic range. A stereo (BTSC) re-modulator cannot be significantly overdriven, so its sensitivity should be set so that a 0 dBFS signal will produce +/-25 kHz p-p deviation 7 kHz (mono mode) for a audio signal of 400Hz at 0dBFS or +/- 12.5 kHz p-p deviation (stereo mode) for a L+R a
35、udio signal of 400Hz at 0dBFS. An AC-3 RF mode sourced signal has speech at -23 dBFS, so will be 6 dB lower in average loudness than a typical NTSC signal. 1With this extra headroom, AC-3 can provide a similar dynamic range to NTSC without the need for the violent non-linear processing commonly empl
36、oyed to avoid over-modulation. 5Copyright Consumer Electronics Association Provided by IHS under license with CEA Not for ResaleNo reproduction or networking permitted without license from IHS-,-,-CEA-CEB-11-A 5.3 MPEG-1 Layer 2 In some equipment, NTSC and ATSC co-exist with MPEG audio. It is theref
37、ore desirable to match its level to the levels discussed in the previous sections. However, MPEG has no formal speech normalization capability, or an operating mode specifically designed to drive an RF re-modulator. Some operating practices control speech so that peaks approach full-scale, others ke
38、ep the speech levels down to allow some headroom for louder sounds, and still others may leave several dB of headroom unused. The following assumes that signals can reach 0 dBFS. Given that the speech level that largely determines the all-over loudness of a program is at about -17 dBFS in an MPEG so
39、urced program signal, and at about -34 dBFS in an AC-3 Line mode sourced program signal, the MPEG signal feeding the analog line output and PCM outputs should be attenuated by 17 dB. The mono re-modulator sensitivity should be set is set so that a 0 dBFS signal drives it to 200% modulation (per sect
40、ion 4.1). An MPEG sourced signal with speech at -17 dBFS will be louder than a typical NTSC signal, but depending on production and operating practices, may be close to the loudness of an NTSC sourced signal. Figures 1 and 2 reflect this. The addition of 6 dB of attenuation in the MPEG signal path i
41、s intended to prevent the stereo BTSC re-modulator from being overdriven. This also drops the loudness 6 dB below that of an NTSC signal, but matches that of an AC-3 RF mode sourced signal. See Figure 3. In all cases, the average level of the program audio sourced by the AC3 decoder in Line mode wil
42、l be lower and not as loud as the signals from the NTSC, MPEG or the AC3 decoder in RF mode. The original, and potentially greater, dynamic range of the Line mode operation makes loudness matching impossible. 6 Line Level Inputs Consumer equipment (CD, MD, DVD players etc.) has a standardized analog
43、 output level (per IEC 61938) of 2 V RMS for a full-scale digital signal. The same standard also requires that the input overload level for equipment shall be no less than 2.8 V RMS (3 dB above 2 V). This suggests that no source should ever be capable of delivering more than a dB or two above 2 V RM
44、S. Note that there are widely varying production practices for media that could be played back via the Line inputs of the Set Top Boxes or Audio Video Receivers discussed in this document, so it is impossible to define a gain structure that will deliver consistent loudness when using the line inputs
45、. 6Copyright Consumer Electronics Association Provided by IHS under license with CEA Not for ResaleNo reproduction or networking permitted without license from IHS-,-,-CEA-CEB11-A 7 Recommendation Summary 2Figures 1, 2 and 3, along with Table 1 provide a summary of the information presented above. T
46、o re-iterate, the gain structure has been formulated so that the average loudness of the input sources will be presented to the listener as uniformly as possible. The main factors determining this are: i) the average (speech) loudness of NTSC programs and of MPEG programs is about 17 dB below 100% m
47、odulation or full scale ii) ii) the average (speech) loudness an AC-3, Line mode sourced signal is about 34 dB below 0 dBFS iii) iii) The RF mode of AC-3 boosts the signal by 11 dB, placing the loudness at about -23 dBFS. iv) iv) a monophonic re-modulator can safely be driven to 200% modulation, whe
48、reas a stereo (BTSC) re-modulator cannot. The default AC-3 decoder operating mode should be RF mode. By defaulting to RF mode, consumer equipment and consumers not prepared for the very large dynamic range available in AC-3 sound broadcasts are provided a basic level of dynamic range without needing
49、 to configure equipment through menus and other methods. Consumers desiring larger dynamic range or the original programming dynamic range on an analog output may switch to Line mode or use digital interfaces such as S/PDIF and HDMI. Source/ operating mode Mono RF modulator relationship* (% modulation) Stereo RF modulator relationship* (Percent modulation) Level in ADC (if present) Baseban
