Differences for MC5.1 at 80kbps with FER

Basic info

• Encoder (float): d18cbee3
• Decoder: 916e6860

Bug description

The item in question is an MLD/SSNR outlier from the selection test verification items. When observing the difference signal, especially for the FER condition, differences in the higher bands are apparent:

MLD is around 1.7 for both clean and FER conditions, but the differences in the signal seem like a bug.

Ways to reproduce

Clean condition

./IVAS_cod_flt_ref -mc 5_1 -max_band FB 80000 48 experiments/basop-verification/MC-5_1/proc_input/bm2aa1s12.wav out/mc-5_1--bm2aa1s12--80000.192
./IVAS_dec_fx 5_1 48 out/mc-5_1--bm2aa1s12--80000.192 out/mc-5_1--bm2aa1s12--80000.clean.dec.5_1.fx.wav
./IVAS_dec_flt_ref 5_1 48 out/mc-5_1--bm2aa1s12--80000.192 out/mc-5_1--bm2aa1s12--80000.clean.dec.5_1.flt.wav

FER condition

./IVAS_cod_flt_ref -mc 5_1 -max_band FB 80000 48 experiments/basop-verification/MC-5_1/proc_input/bm2aa1s12.wav out/mc-5_1--bm2aa1s12--80000.192
eid-xor -vbr -fer out/mc-5_1--bm2aa1s12--80000.192 error_patterns/mc-5_1--bm2aa1s12--80000_ep.192 out/mc-5_1--bm2aa1s12--80000.fer.192
./IVAS_dec_fx 5_1 48 out/mc-5_1--bm2aa1s12--80000.fer.192 out/mc-5_1--bm2aa1s12--80000.fer.dec.5_1.fx.wav
./IVAS_dec_flt_ref 5_1 48 out/mc-5_1--bm2aa1s12--80000.fer.192 out/mc-5_1--bm2aa1s12--80000.fer.dec.5_1.flt.wav