The "tolerance offset" always goes from the allowed absolute maximum.CantusPro wrote: ↑Fri Feb 26, 2021 03:39 CETFirst, -0.95 +/- 0.05 is 1.00 to 0.90 dBTP, and each of my channels are above 0.90 dBTP. That is the quotation from the bottom of the disqualification sheet, and that is how arithmetic works. There may be diverging rules elsewhere, but this is what was written on the bottom of the disqualification sheet.
In case of LUFS, it is +0,3 LU from -16,0 LUFS -- so the allowed maximum offset is -15,7 LUFS
In case of dBTP, it is +0,05 from -1,00 dBTP -- so the allowed maximum offset is -0,95 dBTP
The official rules state - and I quote:
the final mixdown / export must not undershoot -24 LUFS ILk (Integrated), neither exceed -16 LUFS ILk (Integrated), while the maximum digital signal strength must not exceed -1.0 dBTP (True Peak) - NO (Pre)MASTERING. Recommended measurement specifications are ITU-R BS.1770-2+ (ideally ITU-R BS.1770-4) or EBU R-128
The actual allowed calibration tolerances with these meters are +-0,1 LU (standardized) and +-0,02 dBTP.
The offered "tolerance" on the statistic sheet is a "good will" type of situation because of possible offsets from the plethora of signal meters out there in software form. As basis for this "tolerance", I've used ORBAN Loudness Meter (Standalone), which measures both dBTP and LUFS ILk, but only up until ITU-R BS.1770-1 specs, while Wavelab uses the EBU R-128 specs (in fact, it's already EBU R-128 S2 ready, ITU BS.1770-4 is the current equivalent). Due to the missing gating mechanic in ORBAN Loudness Meter, there will be an offset. Else, if I am really in doubt, I am always reaching for Youlean Loudness Meter 2 Pro (budget solution) or Nugen Audio VisLM (high-end solution).
I've explained this in minute detail here:
viewtopic.php?p=8301#p8301 (Post #208)
Doesn't need to be stated, because dBTP, unlike LUFS, is not measured "summed". Each channel is analyzed individually.CantusPro wrote: ↑Fri Feb 26, 2021 03:39 CETSecond, the average of my dBTP levels is -1.15 dBTP ((-1.37+-0.93)/2). It is not specified on the disqualification sheet that one of two channels should not exceed the specified maximum true peak measurement, and the average of my channels has not exceed the specified maximum. There may be diverging rules elsewhere, but this it is not stated on the disqualification sheet.
There is no "average". Sadly, there are meters out there that "sum", which is wrong. A lot of them also offer just one numeric readout for simplification purposes. But in this case, it is always the highest number of one of the channels from all available channels used - no matter if Stereo or up to 11.2 Surround (example).
So in your case, your readout was -1.37 dB (L) / -0.93 dB (R). The highest value would have therefore been -0,93 dBTP. That is +0,07 dBTP higher than the allowed -1,00 dBTP
If it would have been 5.1 surround, and the readouts would have been -2,1 dB (L) / -1,1 dB (C) / -2,3dB (R) / -0,87 dB (LS) / -4,0 dB (RS) / -10 dB (LFE), then the loudest signal would have been -0,87 dBTP max due to an issue in the surround channel, and therefore the stream not being within specs.
I appreciate the effort, but again... the tolerances mentioned in the statistic sheet are a "good will" type situation. The Rule Set still has the highest priority, and that is: do not exceed -16,0 LUFS ILk max and do not exceed -1,00 dBTP max