True Peaks in Audio Mastering Part 2

#007 Unlocking the Enigma 2

Unveiling the Advanced Secrets Behind True Peaks in Audio Mastering – Part 2

Now I can start with the actual blog story for those who sometimes struggle with finding the right balance between our responsibility to do it right and the customers’ desires of loudness-craziness. In fact this is not so easy.

Despite the general advantages of the use of True Peak metering it comes with a set of weaknesses. Many audio engineers are bewildered by interpreting the readings in conjunction with the measured input signal and the desired loudness aims and sonic results. And this is inherent to the weakness of the metric and the complexity of the decision making process.

Fun Fact 1:
Different TP Metering solutions can show different results!
The reason for this is a washy description of the metering standard and an allowed tolerance in the EBU R128 recommendation. Some meters are even outside of the allowed tolerance limits and are technically not allowed to be R128-labeled. When we developed the DRMeter MkII we made sure that it would come with the lowest amount of deviation from the ideal, compared to other major players on the metering market. The EBU website offers testfile downloads which are required to deliver metering results within a defined tolerance. You can google that when you search for “PLOUD EBU test files download” or go here: (https://tech.ebu.ch/groups/ploud).

Fun Fact 2:
Even one and the same meter can spill out slightly varying results on the same source file when you measure multiple times. This is bad but it is system eminent. A solution would be a refined definition of the standard and would require most likely more computation power. The latter is the reason for the flaw. A compromise was needed between precision and CPU-demand.
I often say “metering is a bitch” which is another time true.

True Peaks in Audio Mastering Part 2

A brief guideline for your decision making process when you need to figure out how you set your TP headroom for different versions of a release master:

  1. I recommend my students to keep the “mother master” in the -0.3dB TP range.
    With a little bit of practice and a good personal relationship with your favorite TP limiter you can master the game of nailing the perfect TP structure for your mother master. As a rule of thumb I recommend avoiding consecutive samples at that -0.3dB TP ceiling. But sometimes our customers drive us to a hotter version of the same mother master and you need to play the game a bit harder.
  2. When you go further you need to be aware about the fact that low frequency TP overs are more harmful than high frequency TP overs. Get control of your low end so that it doesn’t trigger TP overs while TP overs approximately above 1k most likely remain acoustically less or not noticeable.
    A few years ago I maintained a research on that. With the help of a little bit of AI I have gathered 400 small audio chunks with the highest peaks of 400 source reference tracks. I gathered a wild set of statistics out of the audio chunks like TP level, number of consecutive top level samples and ratio between consecutive samples to “good” samples. Then I have asked 20 well experienced pro listers to evaluate the sound quality of those files without looking at the wave form. A comparison of the subjective listening evaluation with the factual statistics have led to the conclusion (besides other conclusions) that TP overs which had been triggered e.g. by an acoustic guitar (played by Eric Clapton) in the higher frequency range were completely irrelevant. A beautiful version of his famous song Layla showed TP values of +3.XX dB but the master behaved completely transparent.
    Conclusion: When you need to play the game hard for the questionable aim of more loudness, try to avoid low frequency TP overs while TP overs higher than 1k usually seem to behave transparent; at least in a way that they do not trigger noticeable or problematic distortion. (Side note: Don’t nail me on the 1k transition between problematic and unproblematic. It’s an estimation).
  3. Your customer is king. Being a sound enthusiast is one thing. Serving your clients is another. When a customer wants you to play the loudness game, make sure that you apply the art of getting loud with the least amount of harm (which you learn in our Pro Mastering masterclasses). But probably you may need to just ignore what you see on your TP meter for the sake of making your client happy. Happiness is what counts when it comes to customer satisfaction.
  4. When it comes to your customers deliverables for Online distribution platforms like Spotify and iTunes there are recommendations of having a headroom set to -1dB TP. This is basically the short attempt to mitigate leveling craziness on those platforms and it is not the job of those platforms to educate you to do it right (as we do at Mastering Academy). With a very few exceptions there is no need to play the loudness game hard which is why we teach our mother master principle. Ergo, when you have learned to level perfectly (as briefly described in bullet point 1.), delivering your mother master at -0.3 dB TP is completely fine. They don’t check it. They just want to make it simple and easy to communicate with the -1dB TP headroom recommendation.
    It should not be necessary to mention that those platforms apply loudness normalization based on the metadata. They don’t change the audio itself but they make them being played back with an aligned loudness level based on metadata stored in the audio file. Thus playing the loudness game just makes sense for content which is intended to be downloaded and played by DJs. In such cases it is crucial to play the loudness game for this kind of club music which is promoted to DJs because as soon as the audio file leaves the ecosystem (e.g. Spotify, iTunes) it has the loudness level of your delivered master except some little increase which may be caused by the applied audio data compression format (such as mp3 or AAC).
    A mother master at e.g. -12LUFS would be too quit in loudness compared to the average of about -5 to -7LUFS of the majority of electronic club music releases on Online Platforms to be promoted to DJs, while pop & hip hop works very well between -10 and -12 LUFS.
    Music genres that don’t rely on the compact feel of modern pop music can also work very well between -12 and -16 LUFS and should also be mastered accordingly if more loudness causes damage.
    Good mastering requires a lot of experience and a virtuoso knowledge of the customs and needs of different genres and playout formats. Anyone who is uncertain about this should deal with this topic intensively.

Happy leveling!

Friedemann Tischmeyer

PS: If you like this Post, please don’t forget to share with your friends.