# Basic info - Encoder (float): d18cbee34 - Decoder: 34e965753 (exists also before this) # Bug description As part of verifying fix for issue #996, I noticed that there is a curious very high frequency modulation present in BASOP decoder that is not present in the reference float decoder. Here is a diff spectrogram of `BASOP-ref` generated in Matlab with ``` winsize = 960; figure(1); spectrogram(main-ref, hann(winsize), winsize/2, winsize, 48000); caxis([-150,-70]); colormap(gray); ```  The modulation can be seen at the high end (probably something like 23-24 kHz). This is obviously quite inaudible but can be a clue for some underlying issue. Edit: additional figure highlighting the observed part in spectrogram  # Ways to reproduce Can be reproduced using the instructions in #996, that is, use the switching profile [generic_start_13k2_20frames.bin](/uploads/529174f46529ec934532b85308641812/generic_start_13k2_20frames.bin) with commands ``` ./IVAS_cod -q -masa 2 scripts/testv/stv2MASA2TC48c.met -max_band fb generic_start_13k2_20frames.bin 48 scripts/testv/stv2MASA2TC48c.wav test.bs ./IVAS_dec -q MONO 48 test.bs out.wav ```