Mix Challenge

Sadly, once a company has been picked up, it's not available in the pool anymore. You can select something else, or pass on it. Apologies for the inconvenience.




And really quick regarding your question:

IrionDaRonin wrote: ↑

Fri Aug 20, 2021 15:39 CEST

Anyway, i would like to know what is these 2 points in Loudness/DBTP from the chart. As i would like to improve it in future.

Please check in with the Statistic Sheet (see results post above).

Your track was lower than -14,0 LUFS, but your signal peaked up to -0,08 dBTP (exceeding -1,0 dBTP). This could have been due to the MP3 conversion or something else. Unlike the Mix(ing) Challenge, this is not a disqualification criteria, but merely results in a drop of bonus points. See Rules and Guidelines (post #7), most specifically the Bonus Point mechanic.

Taking a look at the voting results (and keeping in mind that not everyone participated, else the results might have looked different): if you had those 2 extra points, you would have tied with SimaGT. Then it would have been down to who had more higher points after the participant ranking. Assuming a tie, you would have had 4x13pts and 2x12pts, SimaGT would have had 3x13pts and 3x12pts. Since you (Irion) have 1x13pts more, you would have broken the tie and made it to first place.

This was a very, very close run. But sometimes it's all in the bonus points.

Mister Fox wrote: ↑

Fri Aug 20, 2021 05:39 CEST

Ladies and Gentlemen, here are the results of Songwriting Competition 47

I would like to ask how you measure loudness and peaks. I was checking original version and both L and R channel were -1.01 dB true peak. So was not sure so I rather also check uploaded version and redownloaded my own wav track from soundcloud to check and RX8 show I'm in good range. So the question is where is the error

I'm not much into modular synthesis, but maybe it would be nice to learn it and dig it with Cherry Audio Ignite + Year One Collection as prize. Thanks.
I have an account there, in case you need it to put the prize inside my account or something.

Have a good day everyone and thanks for this warm welcome!

Many thanks to everyone for the comments, especially to Mr. Fox. Congratulations to the winners. See you at the next challenge !!

TrojakEW wrote: ↑

Fri Aug 20, 2021 21:16 CEST

I would like to ask how you measure loudness and peaks. I was checking original version and both L and R channel were -1.01 dB true peak. So was not sure so I rather also check uploaded version and redownloaded my own wav track from soundcloud to check and RX8 show I'm in good range. So the question is where is the error

The error is, once more, in the "tolerances".

Personally, I am creating all statistics with Wavelab 10.0.x and it's Loudness Analysis (Batch Processor), since this can export in *.csv, which I can then edit in Open Office for readability purposes. The very same settings as with the Mix(ing) Challenge, only without given tolerances.

Just for understanding what is going on, I ran your file (downloaded on the 21st) through various analysis tools. These are the results:
And this is where tolerances would come into play, as each developer implements their own analysis for True Peak. Some oversample twice, others up to 8 times. Hence why Wavelab has the highest readout (being a high tier mastering tool), why Youlean Loudness Meter goes "red" even though it's -1,0 dBTP, and why RX7/RX8 show -1,01 dBTP. We also have the topic on True Peak save limiters not properly brick-walling at -1,0 dBTP either (depending on the implementation).

I could be more lenient, as with the Mix(ing) Challenge and introduce the same tolerance values - but the question is, where do we draw the line?

You are right with one thing though, I should also mention in this statistic sheet, what the basis for measurements was. Let me handle that with an update in the next couple of hours.

Mister Fox wrote: ↑

Tue Aug 24, 2021 03:44 CEST

The error is, once more, in the "tolerances".

Great. Thanks for explanation. This makes everything even worse since we can trust nobody and nothing, even numbers. So only think I can do is go even for lower true peak values. Lets see what wavelab shows on my 48 entry, but I have to wait for your numbers since I do not have wavelab.

I can already tell you:
You probably know this, but there are many factors involved why there are overshots showing up.

One of them is the different used Oversampling for the dBTP value, depending on the tool in question. And the other is the limiters themselves, promising "accurate True Peak limiting", but still have their own tolerances. This page by "Saintpid Mastering" is probably known to you. Granted, this is an old test (and NUGEN Audio being that "high" with overshots feels a bit off), but it gives you some starting points.


My workaround(s)?

Depending on what limiter I use, which is mostly Barricade 4 (outstanding transparent limiter for low LUFS values), TDL Limiter 6 (or even the old VladG version, both are great for "Pushing" -- only "safe" if you use a compressor/limiter array), or Fabfilter Pro-L... I always set up the max ceiling (threshold) to -1,1dBTP (this was especially important for Barricade 3, as the threshold knob was a bit wonky). If I then use an analysis tool like Youlean Loudness Meter 2 Pro in post, and it lists -1,0 dBTP while staying white/green (doesn't turn red), then things are safe.


I will think about allowing "tolerances" for SWC048... but they might be tighter than for the Mix(ing) Challenge. In fact, I think of "tightening the thumbscrews" for the Mix(ing) Challenge to bring everything in line. Not to mention, Orban Loudness Meter does get a bit long in the tooth and there are plenty of free/affordable more up-to-date alternatives at this point.

Mister Fox wrote: ↑

Tue Aug 24, 2021 17:42 CEST

I can already tell you:

Thank you again. Good info.