Understanding Loudness for Streaming Audio

The concept of "loudness" as it applies to streaming audio is relatively new for many broadcasters. To an even greater degree, it is a foreign concept for content creators and IT specialists charged with managing audio streams who didn't come to the industry through a career in radio and television. To understand what loudness means in the context of streaming audio, a bit of a history lesson is in order.

FM Processing and Loudness

Audio processors have been an integral part of the transmission chain of FM radio stations for many decades. Initially, a simple compressor was used to ensure the audio stayed above the noise floor and a limiter was employed to prevent over-modulation and signal deviation beyond the legal limits.

It is well beyond the scope of this document to explore the admittedly fascinating and often entertaining history of FM processing. Suffice it to say that for broadcast engineers who have been in "the business" for a few years (decades?), the term "loudness" equates to one of the fundamental goals of FM processing and calls to mind the "loudness wars" of the late 1980s through... well, today, in some markets.

Television Audio: Where Loudness Control Began

For those who have worked in the television broadcast world over the last ten years, TV audio went from being something that was just "there" ("Do we have sound? Good.") to paying very close attention to it when listeners rang the switchboard to complain about loud commercials and the FCC got involved.

By June 2009, all full-power analog television in the U.S. had made the transition to ATSC digital (DTV). This change introduced an interesting challenge for the audio portion of the program.

In the NTSC analog format, audio was delivered by an FM carrier alongside the AM visual carrier. While some audio processing was typically used to help manage levels and prevent over-deviation, there were no inherent "loudness" issues with analog TV.

This was not the case with DTV. Despite the fact that the ATSC created a very comprehensive set of recommended practices for digital television audio (known as ATSC A/85 and employing the ITU-R BS.1770 metering recommendations) television broadcasters still struggled to keep levels under control, creating the well-known issue of blaring commercials that were significantly louder than the normal program audio.

This became so problematic and annoyed television viewers so much that in late 2010, the U.S. Congress passed the bill that would become known as the Commercial Advertisement Loudness Mitigation Act, commonly referred to as the "CALM Act." It was adopted by the FCC in late 2011 and officially took effect in late 2012. In short, the CALM Act mandated that U.S. television stations adhere to the recommendations of ATSC A/85, or face repercussions from the FCC if enough viewers complained and a pattern of non-compliance could be established.

Meanwhile, in Europe, the European Broadcasting Union adopted its own recommendation known as EBU R 128. While there are important technical differences between ATSC A/85 and EBU R 128, the underlying goal of both is effective loudness management.

Loudness in Streaming Audio

In the early days of streaming audio, many radio stations streamed their content to the internet, through a built-in player on their own website. Listenership was low, and streaming was treated as a "value-added" proposition by the sales department. Audio levels - so long as they didn't clip the input of the streaming encoder - weren't all that important. A listener would open a stream, adjust the volume, and be done. Many stations didn't bother to process their streaming audio at all, and if they did, they simply dusted off an old analog processor from the back room to provide some basic leveling.

In the past few years, streaming listenership and the number of available streams have increased significantly, elevating it from its previous humble status to one of significance both in terms of the number of listeners and revenue. In short, streaming is now a serious business and is being treated as such by broadcasters and individual content creators.

While most commercial stations still stream to their own websites, they also vie for listeners on dedicated streaming platforms and apps. With listeners now instantaneously switching between streams, variations in loudness have become apparent and annoying.

In an effort to deliver a more enjoyable listening experience for their subscribers, most of the major services delivering on-demand or streaming audio have their own individual requirements regarding loudness. In early 2022, the AES (Audio Engineering Society) released a technical document known as TD1008, "Recommendations for Loudness of Internet Audio Streaming and On-Demand Distribution," to address the subject and provide specific recommendations of its own.

TD1008 is specifically intended for distributors of on-demand audio files and internet audio streams, and we encourage you to read, study, and understand it. It is a relatively brief yet very thorough and thoughtfully-crafted document filled with good information about how loudness is measured, perceived, and managed.

In short - and with the understanding that many streaming audio platforms will have their own unique individual requirements - those wishing to adopt the recommendations outlined in TD1008 should deliver music content at -16 dB LUFS.

Because the spoken word is generally perceived as being comparatively louder when normalized to the same measured LUFS value, speech content should be delivered at -18 dB.

Because it is impractical to separately process music and speech in a real-time environment, TD1008 recommends splitting the difference at -17 LUFS, which falls within the +2 dB LU upper tolerance for the speech recommendation.

Loudness Control in Omnia Forza

Nearly any audio processor - from a simple wideband audio leveler to a full-featured product like Forza - is capable of effectively reducing the dynamic range of incoming audio sources to the extent that listeners won't be annoyed by major differences in levels or sudden loudness shifts.

However, even processed audio can vary in loudness by several dB depending upon the processing algorithms themselves, how they are set up and tuned, and the dynamic characteristics of the source material.

All things being equal, music that was recorded and mastered with heavy compression and limiting and is "dense" by nature will have a higher measurable RMS level and will sound louder than a recording that was more open and dynamic to begin with.

Sometimes these variations are noticeable but not annoying, and sometimes they are quite jarring. In either case, when passed through a LUFS loudness meter using the ITU-R BS.1770 measurement standard, they will usually indicate a measurable difference, sometimes by 3 - 4 dB.

In order to provide the type of longer-term loudness consistency required by most streaming platforms and recommended by AES TD1008, Forza includes an integrated loudness controller that allows users to set both a specific LUFS loudness target and a maximum peak level. It uses an ITU-R BS.1770 meter to measure the loudness of the processed audio and, when necessary, transparently make adjustments to ensure the average loudness levels match the desired target. The loudness controller can be disabled for applications where precise loudness control is not required.