AN AFTERNOON ON LOUDNESS

Size: px

Start display at page:

Download "AN AFTERNOON ON LOUDNESS"

Scot Little
6 years ago
Views:

1 Tuesday 17 April AN AFTERNOON ON LOUDNESS Las Vegas Convention Center, room S204 1 PM Thomas Lund, Dev. Manager HD, TC Electronic: Program Delivery in Accordance with ITU-R BS PM Jay Yeary, Director of Audio & Studio Eng., Turner Studios: Loudness at Turner Studios and Broadcast Engineering 3 PM Florian Camerer, Chairman of EBU P/LOUD: EBU R128 Explained and Implemented 4 PM Roundtable and Refreshments, All Presenters, Q&A: The Welcome Revolution

2 Program Delivery In Accordance With BS Thomas Lund TC Electronic A/S Denmark NAB Afternoon on Loudness Las Vegas April 17, 2012

3 Agenda The Standards From BS.1770 to BS Loudness, True-peak, Gating Production Guidelines, EBU R128 Tools Loudness Meter Harmonization Transmission Worldview Distribution, Cross-platform, Commercials, Metadata ATSC A/85, TR-B32, OP-59, EBU R128 BS and BS

6 Loudness

7 Leq Weighting K

8 BS.1770 Loudness Power Gain Sum Left Pre Filter RLB Filter Mean Square 0.0 db Right Center L Srnd R Srnd LFE Pre Filter RLB Filter Mean Square 0.0 db Pre Filter RLB Filter Mean Square 0.0 db Pre Filter RLB Filter Mean Square +1.5 db Pre Filter RLB Filter Mean Square +1.5 db Currently not used in Loudness Calculations Works for mono, stereo and 5.1. Digital input preferable

9 Words: LUFS, LKFS and LU LUFS = LKFS Unit for absolute loudness level Compare with [dbfs] for peak level LUFS is the same as LKFS Only the spelling is different LU Unit for relative loudness level Compare with [db] for peak level Loudness at Target Level = 0 LU

10 LUFS and LU 0 LU denotes Target Level. In this example, Target Level is -23 LUFS. Remember: LUFS is the same as LKFS.

11 Meter Examples: LUFS and LU

12 Program Loudness LUFS Use to normalize programs A static gain of +5 db brings Program Loudness on Target Ingest Corrected

13 Speech and all that Jazz - Dialog normalization is easily fooled, and creates leveljumps between programs and commercials. - Mixing esthetics are different from genre to genre. - Measuring speech is ambiguous and patent protected. - There are better alternatives...

14 Speech normalized example Four Programs: 1. Movie 2. Commercial 3. Drama 4. Pop = Too simplistic LUFS Broadcast is not Film (speech) Broadcast is not Music (peak)

15 News Speaker (NLR) Green: Ungated Red: Gated LUFS

16 Episode of Friends (MLR) Green: Ungated Red: Gated LUFS

17 The Matrix Movie (WLR) Red: Gated Green: Ungated LUFS

18 Measurement Gate EBU A relative gate between -10 and -6 LU gave significantly better normalization results than no gate, or other gating schemes. NHK & ARIB Independent verification by NAB/ARIB and by NHK support relative gating method and a threshold of -8 LU. CRC Independent verification: -8 LU relative gate shows best MSD performance.

20 CRC

21 Leq(K) Loudness Un-gated ATSC A/ LU BS , R128, TR-B32 Includes silence Application critical Cross-genre critical Focus on foreground Application friendly Cross-genre friendly Similar numbers with NLR material Different numbers with WLR material

22 True-peak Level AES 23 Nielsen & Lund, 2003 Even in a linear audio system, analog and digital level is not the same. Analog Level: Black Line Digital Level: Red Dots 1 2

23 True-peak Level AES 23 Nielsen & Lund, 2003 Even in a linear audio system, analog and digital level is not the same. Analog Level: Black Line Digital Level: Red Dots 1 True-peak Level: Grey Lines 2

24 CD True-peak Level Original CD level Bit cloned No gain change No sample rate conversion Note: Meter must be digital & synchronous in order to display true-peak level correctly

25 0 dbfs+ Listening example: Don t Stop, Anastacia Original CD, normal stereo Remember how loud this track is in order to compare against distortion level in the next examples.

26 0 dbfs+ Same song, NAD512 CD player, listening here

27 Data Reduction AES 23, 2003

Data Bandwidth 50% growth p.a. = double in 20 months 1995 1996 1997

28 Data Bandwidth 50% growth p.a. = double in 20 months

29 Agenda The Standards From BS.1770 to BS Loudness, True-peak, Gating Production Guidelines, EBU R128 Tools Loudness Meter Harmonization Transmission Worldview Distribution, Cross-platform, Commercials, Metadata ATSC A/85, TR-B32, OP-59, EBU R128 BS and BS

30 production logging ingest transmission

31 R BS.1770 THE UTION LOVE

32 Loudness on three time-scales Momentary Loudness Short-term Loudness Program Loudness

33 Harmonized Metering Momentary Loudness M 400 ms measure of loudness. Un-gated rectangular window. Short-term Loudness S 3 sec measure of loudness. Un-gated sliding window. Program Loudness I The loudness of a full program. Gated measure. Same as BS Loudness Range LRA The loudness variations of a program. Independent of absolute level.

34 Loudness Range Example: LRA = 10 LU To keep a constant loudness during the program, a +- 5 db fader gain ride would be required.

35 Loudness Range Example: LRA = 10 LU To keep a constant loudness during the program, a +- 5 db fader gain ride would be required. Fader ride

36 Loudness Range Example: LRA = 10 LU To keep a constant loudness during the program, a +- 5 db fader gain ride would be required. Fader ride

37 Loudness Range in Production Delivery Specs Guideline: Settle expectations. Not car radio, not cinema.

38 EBU Mode Features A set of requirements for loudness meters How loudness is measured, BS.1770 What numbers are displayed Not the design of meters => different meters, same numbers

39 EBU Meter Example Target Level Momentary Loudness 400 ms, M Short-term Loudness 3 sec, S 6 LU Target Level At 12 o clock and at bold circle Loudness Range Variation of program, LRA Program Loudness Loudness of program, I

40 Live Metering Loudness meters easier to use than peak or VU Look at a Sliding measure and Program Loudness Start 1 LU below the Target Tolerance allowed Peak meters are used only to avoid overload Move up to the Target at the end

41 Live Sports Sliding Window and Program Loudness

42 Production Guidelines Speech Regular speech: -27 to -23 LUFS Music Foreground: -24 to -21 LUFS Loudness Range Get expectations cleared early Note: Also useful for speech only programs True-peak Level No higher than -1 dbtp May need further lowering downstream, depending on platform and codec

43 coming up: Mixing example using BS normalization Monty Croc Olympics LUFS => gain -1.7 db Rock n Roll Train, AC/DC -5.7 LUFS => gain db (!) Ruby Baby, Donald Fagen LUFS => gain -3.3 db BBC Jingle LUFS => gain -7.5 db Monty Croc Olympics (same) BBC Jingle (same)

44 Production Summary Peak Level is not important, unless -1 dbtp is exceeded Use Program Loudness to target -23/-24 LUFS Use Loudness Range (LRA) as production guideline Observe Momentary (M) and Short-term (S) Loudness Consider mixing at a calibrated listening level

45 Agenda The Standards From BS.1770 to BS Loudness, True-peak, Gating Production Guidelines, EBU R128 Tools Loudness Meter Harmonization Transmission Worldview Distribution, Cross-platform, Commercials, Metadata ATSC A/85, TR-B32, OP-59, EBU R128 BS and BS

46 Loudness Jumps How much can be tolerated between programs? 2003 Comfort Zone paper - Blind to anything but speech - Study not based on BS.1770 when measuring segments AES 2009 paper: New study based on BS Commercials, drama, news, pop music, movie - Measures all sources

47 Loudness Jump Tolerance Tests with mostly speech Tests with mostly music and effects

48 Loudness Jump Tolerance Results 50% would adjust volume for a loudness increase of 3 LU for a loudness decrease of 6 LU 95% would adjust volume for a loudness increase of 5 LU for a loudness decrease of 8 LU Conclusion Any type of sound could trigger a level-adjustment.

49 Where to Normalize At the station or using metadata? EBU recommends normalization at the station: - one thing less to go wrong - a benefit to all platforms - more predictable - lower workload - cheaper (fewer Dolby boxes) Normalization => Less transmission processing itunes example

50 Loudness Range in Transmission Delivery Specs Guideline: Settle expectations. Not FM radio, not cinema. Compatibility Loudness Range predicts if a program fits consumer requirements, or if platform processing is needed. Quality Control To check downstream signal-path incl. satellite, cable and STB.

51 Loudness Range The ideal loudness range depends on the listener. <20 LU Target Loudness Range <12 LU <8 LU <6 LU A high background noise calls for a low loudness range.

52 Program Loudness at -24 LUFS is too low for pod/mobile platforms Apple Sound Check aims at LUFS using BS no G G8 Max LRA TP Broadcast Pop, Rock, Clas. Music

53 Platform Loudness LU -1 dbfs -3-3 dbfs Trickle-down: +18 Mobile Platforms dbfs LUFS 1. Normalize 2. LRA and Limit 3. Make-up gain Note: 0 LU = -24 LUFS LUFS is the same as LKFS HD Production Mobile Platforms Mobile Platforms

54 Getting the best out of AC3 Normalize Limit Metadata Normalize programs at the station using BS Protect AC3 from peak overload, especially when transmitting 5.1 format Change metadata only when switching between stereo and 5.1 Enter -23/-24 as dialnorm number Note: The AC3 decoder is not speech based. Dialnorm just tells the average level of a program

55 Agenda The Standards From BS.1770 to BS Loudness, True-peak, Gating Production Guidelines, EBU R128 Tools Loudness Meter Harmonization Transmission Worldview Distribution, Cross-platform, Commercials, Metadata ATSC A/85, TR-B32, OP-59, EBU R128 BS and BS

56 Headroom For WLR content, BS provides 3-4 db more headroom than BS Bolero, Matrix, Pirates etc. For movies, BS provides 2-10 db more headroom than Dolby speech gating

57 BS Use to normalize programs at the station to -23 or -24 LUFS => more headroom than we have ever had in broadcast!

58 Common understanding: Loudness instead of Peak level A/85 R128 HDTV Only Focus Anchor for WLR All Platforms One Number Monitoring guidelines AC3 manual Distribution guidelines Open standards Harmonized metering Tools to control commercials

59 Where is ATSC on Loudness?

60 One commercial Two results The relative gate enables the measure to focus on foreground sound: Louder and potentially annoying parts of a program. Loudness 0 LUFS Target BS BS

61 Hot Commercials 2nd line of defense: Max Short-term Loudness 0 LUFS Max Target Max Short-term (S) prevents programs from ever getting overly loud

62 Transparent rules Transparency when normalizing commercials is a must - for the advertiser - for the production house - for the broadcaster

63 BS Better at controlling commercials Better at normalizing WLR content Better at setting metadata than a speech or a peak based measure Enables transparent normalization across all platforms Open standard, no proprietary technology Automated handling, runs on a server Harmonized metering across brands => cheaper

64 International Status Feb 2012 Europe: EBU R128 UK, France, Norway, Italy, Israel already started. Germany, Austria, Switzerland: August, 2012 Rest of Scandinavia, Poland, Netherlands: During 2012 Japan, Brazil Already using BS , LRA, M and S measures China BS starting now, incl. M and S measures United States: ATSC A/85 CALM Act will be enforced from Dec, Commercials: (Presumably) old BS.1770 with no gate. Programs: Speech level.

65 The War Is Over Hello group! Feb 9, 2012, 20:14:44 GMT The radio loudness war in Norway is over! I'm happy to inform you that the DAB and DAB+ platforms in Norway as of today are loudness normalized! In a meeting at Lillehammer yesterday all Norwegian DAB/DAB+ content providers agreed to call off the loudness war and back down to a common target level. We are not at -23 LUFS yet, instead we chose -15 LUFS as a preliminary target level. The main reason for this choice is to match the levels between DAB/FM/WiFi in multiplatform receivers. The plan is to shut down FM in Norway in Then we can lower the target level to -23 LUFS, if not before. Please drop me an if you want more information. Regards, Bjørn Aarseth Norwegian Broadcasting Corp. Technology Department R&D Bjørn Aarseth, Senior Engineer Tel. direct: Mobile: Postal address: TBLK, FN31, 0340 Oslo, Norway

66 Ugly CD Examples LUFS PROGRAM LOUDNESS 1. Danish Pop 2. Kelly Clarkson 3. Metallica 4. Pink Noise (FS) LUFS PROGRAM LOUDNESS Don t let the same happen to broadcast

67 References Read More Zwicker & Fastl, 1990 Psychoacoustics - Facts and Models Moore et al., 2003 (JAES no 12) Why are Commercials so Loud? Nielsen & Lund, (AES 23, 107, 109, 111, 121) 0 dbfs+ Level in Mastering and Audio Production Skovenborg & Lund, (AES 123, 125, 127) Loudness Descriptors to Characterize Programs and Music Tracks ITU-R BS , ATSC A/85, TR-B32, EBU R128, Tech docs Glossary: IRT NoTube project: End AES 132 Budapest Convention, April 27 Loudness Range - Design and Evaluation Thomas Lund, thomasl@tcelectronic.com

Normalization of Broadcast Audio

Normalization of Broadcast Audio T h o m a s L u n d T C E l e c t r o n i c A / S D e n m a r k To n m e i s t e r t a g u n g L e i p z i g N o v 2 7, 2 0 1 0 Peak Level Even in a linear audio system,