OK, thanks for the files, I see it, strange. I will analyze it more.
Tracked as: https://www.ventismedia.com/mantis/view.php?id=18717

And this shows the situation in one single picture:
[url]https://e1.pcloud.link/publink/show?code=XZiahFZWPsO1q62ygLwKYv67pJ0wzgYfeaV[/url]

Thank you for taking care of this problem, but unfortunately I have to say that I actually did lose also sound samples. Under the following link you can find five files: [url]https://e1.pcloud.link/publink/show?code=kZyyhFZFihXj6SL5kj5cWBdbjwrzSoIl3ik[/url]
One is an original MP3 file (under the download link probably displayed as picture) and four are conversions starting from that. There are two OGG files and two WAV files. The files labeled MM were converted with MediaMonkey and the files labeled VP with VideoProc Converter. If you load these files in a DAW you can see that the ones converted with MediaMonkey are missing samples at the beginning of the track, including sound samples (in this case the sound content begins at ~2 ms already). The VP labeled files, in contrast, resemble the orignal MP3 completely.

Yes it slightly varies, but it should be always related only to the tiny alignment gaps (few ms), it should not affect audio data (you are not losing samples with sound, only samples with silent). It is related to MP3 decoding, not OGG encoding (you can experience the same with saving to WAV too, i.e. without encoding).
I tried another decoders and SW and they behave the same, only the gap could by slightly different.

After examining some more files, I can say that the cut of 528 samples does not seem to be consistent. In some cases also samples are added at the beginning of files. From my understanding of the OGG format, however, the number of samples of the original file should not be changed at all, and there should be definitely no chance that sound information is lost which happened to some files I converted.

Not sure if this effect depends on my windows system or if it's inherent to MM.

I think MP3 and OGG are both lossy formats, MP3 having the property to sometimes add null samples when conversions into MP3 are performed. However, I do not expect to lose samples by any kind of conversion. Viewing some of my files in a DAW revealed that the MM algorithm always loses exactly 528 samples per track (checked for conversions to OGG and FLAC), also when audio material is contained. While normally MM is my preferred software for conversion, this time I cross-checked with a different converter where in fact all samples were preserved. As far as I know the other converter employed uses Ffmpeg algorithms.

This is normal, MP3 is lossy format and nearly always has some small gap at the beginning. Decoders, like the one included in Windows used by MM, then do not decode always the whole gap before audio. It should not affect non-gap audio, do you experience some problem, where also some audio is cut?

When converting files from MP3 to OGG MediaMonkey always cuts off a certain number of samples from the beginning of each track (exactly 528 samples at sample rate 44.1 kHz). I observed this effect for conversions done with variable as well as with fixed bitrate, for MM versions 2433 and the new 2531. The said problem does not seem to occur when starting from other formats; FLAC to OGG conversion for example delivers complete destination files.

Maybe someone can verify this.