I was messing about with the TestGenerator when I noticed that the peak output level for pink noise was higher than it should be by +2.5dB. All the sine, triangle, square, sawtooth,white noise and Brownian noise levels peaked at the level value set within TestGenerator, but for pink noise, the peak value was consistently higher than the output level setting e.g. set to -20dB it peaked at -17.5dB and -30dB peaked at -27.5dB.
Note that I had already set the panning law to 0dB, so that there was no reduction in the signal level from the channel to take into consideration.
Whilst the effect can easily be compensated for (if known about), it would be nice if this small error could be fixed. I couldn’t see this issue raised anywhere else.