Stereo Tool

I hope that it is not me, who has misunderstood both (Hans and Bojcha)

Ok, some more output data then! Based on 128 kbit MP3.

Slightly increasing the pre-emphasis does not really protect against every type of clipping, but it does protect loud 'S' sounds - which can sound really bad when clipping occurs.

Here are some frequency analyses:

Output level 100% (Post Amp = 1.00):


Output level 90% (Post Amp = 0.90, -1 dB):


Starting at 20 us, the pre-emphasis starts to have a noticeable effect on the 'S' sounds - so I would advice against using values higher than 20. That's with a Loudness setting of 3.60 (very high) - at lower Loudness settings the effect is much smaller (higher loudness setting = more RMS pressure = more difficult to handle; also more energy in the high frequencies (due to higher overall audio level) = bigger effect of the pre-emphasis).

In both images at 20 us the amount of clipping distortion (the lines shown above ~ 16 kHz) is greatly reduced. So I guess around 15-25 us (depending on encoding quality etc.) could be a good value.

At output level 90%, the number of clipped samples for the same 30 seconds of audio is:
OFF: 3500
0 us: 3300 (should have been the same... Probably just a slightly different portion of audio?)
10 us: 2300
20 us: 900
30 us: 300
40 us: 50
50 us: 20
So the big jump occurs between 10 and 20 us.

Clipping distortion above 18 kHz:
OFF: -63.8 dB
0 us: -64.0 dB
10 us: -66.8 dB
20 us: -73.2 dB
30 us: -81.0 dB
40 us: -90.0 dB
50 us: -91.5 dB
The biggest jump in absolute numbers (and I expect by far the biggest effect in audio quality) is between 10 and 20, but after 20 it keeps being reduced at a very high rate.

Idea for a new feature: Less strict limiting for web radio stations.

Rationale: For web radio stations, it's not really useful to limit the sound very strictly below a peak level, because the peaks change anyway during the MP3/AAC/... encoding.

In some cases (difficult sound) the very strict filtering DOES have a pretty big effect on audio quality.

So: For web radio, turn off the final filtering stage (that removes all the remaining peaks). "HARD LIMIT" should be turned off as well, but this idea also includes skipping the last stage of the "Loudness" filter.

Bojcha, as far as I understood from you, you determined the levels at 192 kHz. Then later you reported a bug: AGC speed at different sample rates is different. That bug has been fixed, now everything behaves as 44 kHz did before. So your levels determined at 192 kHz should now be divided by 4 to get the same effect at 192 kHz.

My current speeds are Up ~0.80 and down ~10.00.

HANS IF YOU GET THIS I SENT YOU 30 euros

Idea for a new feature: Less strict limiting for web radio stations.

Rationale: For web radio stations, it's not really useful to limit the sound very strictly below a peak level, because the peaks change anyway during the MP3/AAC/... encoding.

In some cases (difficult sound) the very strict filtering DOES have a pretty big effect on audio quality.

So: For web radio, turn off the final filtering stage (that removes all the remaining peaks). "HARD LIMIT" should be turned off as well, but this idea also includes skipping the last stage of the "Loudness" filter.

My current speeds are Up ~0.80 and down ~10.00.