[HELP] Hybrid introducing audio delay that is not present in source material

[Selur]

I did a quick test with 5.1 audio, MP4Box and muxer:
(assuming VirtualDub2 properly loads and displays the audio; if this is wrong everything that follows is false)
Original:

encoded with FDK:

Encoded with FAAC:

Encoded with FHG

Encoded with Nero:

Encoded with QAAC:


So to me it looks like:
Original vs FAAC_no_encoder_compensation; audio moved half a frame backward
Original vs FDK_no_encoder_compensation: no delay
Original vs. FHG_no_encoder_compensation: audio moved ~3/4 frame backwards
Original vs Nero_no_encoder_compensation: audio moved one and a half frame backwards
Original vs QAAC_no_encoder_compensation: audio moved one and a half frame backwards

Would you agree with these findings based on the images?
Can you reproduce these findings (I attached the source)?

Assuming the findings are correct, then:
a. the only really good delay handling happens with FDK_no_encoder_compensation.
b. yes, in encoder compensation is needed since using no_encoder_compensation isn't correct.

---
General problem:
a. no clue how accurate VirtualDub2 is (for exact adjustments a single ms accurate audio wave form would be needed)
b. to adjust the encoder delay a ton of extensive testing needs to be done so that the delay can be adjusted for each decoder depending on sampling rate, encoding mode and may be channel count. + muxer (FFmpeg/MP4Box and container mp4/mkv/...) + whether or not the input already has a delay.

Cu Selur

Ps.: Note that a delay of ~1 frame is usually not visible during casual playback this is only reall interessting if you encode to aac and plan to use the output as a editing base for later projects.
Even experts have can't usually see (or more like feel) this unless they are really familiar with the source. Singers sometimes spot the problem when they see frame by frame clips with underlayed audio waves.
So this isn't a really concerning matter.