Program suppliers and broadcasters alike have long been plagued by surprise level changes when switching from one source to another. The volume, or loudness, can vary wildly from source to source, wreaking havoc with unattended operation.
Significant differences in loudness often exist between television programming and advertisements. For viewers, this is irritating because it results in a constant need to reach for the remote and adjust the volume. The problem isn't just confined to commercials. Notable loudness differences can also exist in the same program, for example with live sports feeds or during a movie or a talk show.
Surprise level changes are also an issue that rebroadcasters have to deal with, especially satellite headend or IPTV providers, who have absolutely no control over the broadcasted content's average operating levels.
All of these facts raise various questions: Are the loudest programs really the best? Is loudness any guarantee of quality? And why worry about it when this is something broadcasters and viewers have been dealing with for years?
The problem is that nowadays, with a much broader content offering and so many different viewing and listening tools available (television, Internet, mobile devices, DVDs), what used to be an annoying side effect has become a major concern. Audiences are tired of watching TV with the remote control glued to their hands, and the program makers and presenters are equally unhappy. Complaints are coming in from everywhere, to the extent that even governments are getting involved by considering legislation to address this issue. Broadcasters and network operators are not only being blamed for the problem but also being asked to sort it out.
What is actually happening?
Broadcast playout centers receive radio and TV programs from an incoming source and then distribute them on to the viewer. Sometimes, prior to analog transmission, the sources are modulated and the frequency converted into different channels. In the digital environment, sources are received and decoded before new encoding and multiplexing is carried out prior to further distribution.
With the migration to HDTV services, the problem of loudness becomes even more relevant as the audio for most digital HD services is transmitted in multichannel coded format. The most common standard used is the Dolby Digital multichannel digital audio coding format. Today we often see a remarkable difference in loudness between pulse code modulation (PCM) stereo tracks and related 5.1 coded Dolby Digital streams.
To measure what is happening to the audio signal, broadcasters have usually relied on PPM and quasi-PPM. But when you are dealing with loudness, it is necessary to introduce another measurement: loudness units. Figure 1 on page 42 shows why this is important.
In the upper graph on Figure 1, the audio stream from various programs has not been controlled by any loudness device. The loudness level jumps significantly between programs (represented by the dark blue line). The lower graph shows what happens to the same audio stream when the broadcaster has introduced a loudness control device and can measure in loudness units. The dark blue line becomes much flatter, indicating that the loudness level between the various programs is more consistent, making it much easier on the listener's ear.
To control loudness, it is necessary to perform some kind of gain control to level the volume between the various audio sources. Yet, to make this acceptable to the viewer, the gain control needs to be inaudible. To understand this, we need to look at the psychoacoustic effects.
The human ear does not recognize changes in the intensity of an acoustical event, provided the transition time is long enough. If a gain change is performed at 1dB in three to four seconds, the human ear is physically incapable of noticing that change. This means that if we perform gain changes at a speed that is no faster than 1dB per three seconds, the manipulation will not be perceptible to the listener. We can achieve inaudible automated gain control as long as the circuit doesn't try to perform gain control too quickly.
There is another psychoacoustic effect that is related to immediate audio level changes. In principal, listening to dynamic sound isn't irritating. In fact, we need to perceive different audio levels in order to appreciate the dramatic content of a program. For example, high audio levels heighten excitement and suspense, while lower levels might be used to convey emotions or carry dialogue. If the audio level drops or jumps more than 12dB to 14dB unpredictably from one moment to another, it becomes difficult for our senses to comprehend. For the listener, these unexpected and immediate loudness changes can be unpleasant.
What is the solution?
Loudness needs to be handled differently and can't just be measured using peak level. With this in mind, Jünger Audio developed Level Magic, a comprehensive series of level control devices that successfully addresses all of the issues broadcasters currently face, both with stereo and 5.1, and with audio from a wide variety of sources.
The system uses an adaptive loudness control algorithm that simultaneously involves three basic circuit designs. These include an adaptive AGC, a transient processor and a distortion-free brick wall limiter. For the leveler (AGC and transient processor), Jünger Audio uses a unique combination of level detectors to instantly analyze the incoming audio signal. Dependent on the analysis, the resulting gain change is applied. Two gain change elements together are working in the chain — the transient processor as the fast gain change element and the AGC as the slow gain control element.
As already discussed, it is important to perform the AGC slowly so that the resulting gain change stays inaudible at all times. However, setting the AGC in this way makes it difficult to deal with transients and fast incoming level changes. The transient processor is incorporated into the signal path, thus allowing the system to control transient elements of the audio signal.
The proprietary algorithms in the system also allow the automatic adjustment of the attack and release times according to the evolution of the input signal over time. This is called adaptive processing control. By monitoring the waveform of the incoming audio, the system can set relatively long attack times during steady-state signal conditions and short attack times when there are impulsive transients.
The digital implementation of the multiloop design also permits a short time delay to be introduced in the audio signal path. This allows the gain changing elements to look ahead and determine the correction needed, and then apply it to the delayed signal to control even the fastest transients. This is particularly important for the limiter, which provides a precisely leveled output signal free of overshoots (clipping) and distortion.
The adaptive controlled multiloop algorithm makes it possible to create sophisticated audio level control solutions that need little input from the end user. The system's algorithm offers a “set and forget” application. Overall, the signal loudness is under control without artifacts previously associated with such systems. And because the processor is not multiband, the spectral balance is not touched.
Earlier this year, Jünger Audio introduced an ITU 1770-based level detection algorithm into all broadcast versions of the Level Magic hardware processor boxes. As a result, the product not only creates consistent audio loudness, but also it complies with the ITU's loudness control specifications as level detection can easily be switched to ITU mode.
The C8000 modular hardware system incorporates 5.1 and stereo audio level control compliant to ITU 1770, AES/EBU interfacing and HD/SD-SDI embedding, a range of Dolby hardware modules, and an optional metadata generator. The metadata generator allows related Dolby metadata to be verified, set and transferred internally to the encoder circuit for use in the coded stream. All of the cards are configured by a Web interface and can be monitored by a management system using the SNMP option of the LAN controller card.
The absence of a loudness reference is certainly creating problems, not to mention heavy debate among industry gurus as to whether we should all follow the Dolby flag, which targets a reference value for loudness at -31dBFS, or whether to find a consensus at around -20dBFS, which is the figure presently being discussed by the EBU's P-LOUD group.
While no one knows what the final standards for loudness will be, Jünger Audio is preparing answers for all options — and recognizing that the loudness issue will always have to be handled somehow.
Peter Pörs is managing director at Jünger Audio.