Clipping is the distortion.
It can happen when you downmix the channels to different configuration.
7.1 to
5.1/2.0 or 5.1 to 2.0, etc.
Usually this is because the audio is mastered at very high levels so when the channels are mixed together, the combined loudness will be too much for a single channel (back left + front left + half center into one channel: front left) and cause this distortion.
Dialogue won't be clipped most of the time. SFX will sometimes be noticably clipped.
Music will always be noticable since it runs for longer periods of times and can vary widely in volume.
Clipping mostly affects bass and produces ugly bass distortion.
I'm not making any spectrums to show what a clipped audio waveform looks like, you can google on it.
"float" is the type of bit depth, but only for bit depths 32 and above.
It's math. Floating type vs. integer (32.88089 the values after the decimal point make it a floating type VS. 32.0000 the constant zero values after the decimal point make it a solid integer).
16 bit (normal bit depth for CD audio; sometimes BD audio; lossless digital music sales) is always integer.
24 bit (normal bit depth for DVD/HD-DVD/BD audio formats; rarely lossless digital music sales) is always integer.
32 bit (common bit depth for audio processing/recording for home users) can be either integer or float point.
If we select float point, we will never see the actual value. We don't need to. Ever.
It's just to mathematically help with any audio processing (Equilizer, channel mixing, etc).
Using 24bit and higher will help reduce this clipping distortion.
Using 32, you can choose float point (for more accuracy) but your soundcard might not like using floating point and cause extra noise or even more distortion.
Using 32 integer will most likely be safer than floating point if you don't know whether or not your soundcard won't throw a fit over floating values.
Really, once you use 24 bit, everything is just overkill beyond that value.
This is rather odd, then. Your settings shouldn't have any problem.
But, why your DVD has such terrible corruption on downmixing to stereo is rather odd.
Is this a retail bought disc or some encode? Is it Region 1?
I'm going to encode the blu-ray audio down to DVD-compliant AC3 (448kbps/6ch = 74.66666666666667kbps per channel) and then give that a go to see if it clips when downmixed to stereo.
I'm also going to do a stereo encode that is DVD-compliant AC3 (448kbps/2ch = 224kbps per channel).
However, the fact that it plays fine when left in original 5.1 audio and corrupts when downmixing to stereo is enough to prove that something's not downmixing properly.
It could be just the DVD's have not been normalized to take downmixing into consideration.
The last DVD I bought was... Dexter season 8. On account the rest of my Dexter seasons were also in DVD.
---------- Post added at 05:32 AM ---------- Previous post was at 05:30 AM ----------
oh wait... you're from Belgium. Is that where you're really from?
Region 2? 25 frames per second?