Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Significant difference between true peak on/off (Read 1149 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Significant difference between true peak on/off

I normally scan my albums with true peak turned on and set to auto 8x oversample. Usually the difference between true peak on and off (have tested a few times) is in the second/third decimal place or lower.

However I have a few albums which were digitally released as 24-bit/44.1 KHz. With these the difference is significant. For example:
album peak (true peak 8x): 1.145963
album peak (true peak off): 0.959510

What's going on and which value should I trust? I haven't changed the default setting to downsample high-definition content but I don't think that would apply here anyways.

Re: Significant difference between true peak on/off

Reply #1 –
App: Foobar2000 2.1 built on 2023-12-11
Track: Vangelis - Juno's quiet determination.flac (44 kHz, 16 bit)

Track peak valueSpeed xTrue peak scan method
0.710266351.19none (Replaygain out-of-the-box)
0.724841177.16auto 2x
0.724841129.73auto 4x
0.72484174.36auto 8x
0.711239119.17RetroArch 96000 Hz Normal
0.71053719.01RetroArch 96000 Hz Highest
0.71051951.18SoX 96000 Hz Normal
0.71051748.05SoX 96000 Hz High
0.71051012.78SSRC 96000 Hz


The results do raise questions indeed. Why are the values calculated in automatic mode noticeably different from the values of prominent resamplers (0.724 vs 0.710)? And if the latter almost equal the values without oversampling (0.710... = 0.710....), then is it worth using True peak scan, bearing in mind a noticeable loss in calculation speed (351x vs 51x)?
• Join our efforts to make Helix MP3 encoder great again
• Opus complexity & qAAC dependence on Apple is an aberration from Vorbis & Musepack breakthroughs
• Let's pray that D. Bryant improve WavPack hybrid, C. Helmrich update FSLAC, M. van Beurden teach FLAC to handle non-audio data

Re: Significant difference between true peak on/off

Reply #2 –
Quote from: gfxnow on
What's going on
Different signals have different impact. Hadn't it been so, one could just have put it into a one-size-fits-all formula.

There are many threads at HA about these. A recent one with links to others: https://hydrogenaud.io/index.php/topic,124977


Quote from: gfxnow on
and which value should I trust?
Typically the highest. It doesn't matter as long as you don't hit a risk of clipping. Typical playback mode is to adjust volume according to the gain tag or peak tag depending on which one gives the lowest volume - and in most cases that is the gain tag, and by so much that a couple of dB on the peak wouldn't matter. But the impact could be higher than what you quoted (a ratio in the 1.5 to 2 dB ballpark) on other signals.

So if you play back with gain adjustment and prevent clipping according to peak, then tagging with the highest value has zero consequences for most files, and when it has some ... that is when you might want it.
The "adverse consequence" of a more accurate true-peak scan, is on your CPU time.

Note, if you play with the volume slider not-all-the-way to the right, it would also prevent.

Re: Significant difference between true peak on/off

Reply #3 –
Out of curiosity, I asked fb2k to generate a 1kHz tone and tried a similar test. The tone appears to be 44.1kHz. At least that was the resulting sample rate when I converted it to a wave file. I assume in a perfect world we're looking for a value of 1.000000 for true peak scanning (or would it be 0.999999?), so I guess resampling is something of an art-form.

Track peak valueTrue peak scan method
0.999994none
1.018921fb2k auto - 2x / 4x / 8x
1.000058RetroArch - 88200 Hz / 176400 Hz - Highest
0.999990RetroArch - 96000 Hz / 192000 Hz - Highest
0.999994SoX - 88200 Hz / 176400 Hz - Best
1.000000SoX - 96000 Hz / 192000 Hz - Best
1.000064Spex - 88200 Hz/ 176400 Hz - Max
1.000163Spex - 96000 Hz / 192000 Hz - Max
1.000103dBpoweramp SSRC - 88200 Hz
1.000193dBpoweramp SSRC - 96000 Hz
1.000055 / 0.999994SRC - 88200 Hz / 176400 Hz - Best
1.000022 / 0.999974SRC - 96000 Hz / 192000 Hz - Best

Re: Significant difference between true peak on/off

Reply #4 –
Quote from: yetanotherid on
Out of curiosity, I asked fb2k to generate a 1kHz tone and tried a similar test. The tone appears to be 44.1kHz. At least that was the resulting sample rate when I converted it to a wave file. I assume in a perfect world we're looking for a value of 1.000000 for true peak scanning (or would it be 0.999999?), so I guess resampling is something of an art-form.
Try to generate a 20kHz tone for funnier results.