Large difference between BASOP and float for object editing test case with JBM

Basic info

• Float reference:

  Encoder (float):

  • Decoder (float):

• Fixed point:

  • Encoder (fixed):
  • Decoder (fixed):

Bug description

While working on !2415 (merged), we noticed a regression that appears to be an existing separate issue related to object editing.

 Below is the plot of the float and basop output, where it’s clear that the basop output level intermittently drops and then returns.

The test case is:

test_param_file_tests[stv-3 ISM with metadata at 384 kbps, 48 kHz in, 48 kHz out, FOA out, object editing, JBM Prof 5, DISC]

IVAS_cod -ism 3 scripts/testv/stvISM1.csv scripts/testv/stvISM2.csv scripts/testv/stvISM3.csv 384000 48 scripts/testv/stv3ISM48s.wav stv3ISM48s.wav_3_ISM_with_metadata_at_384_kbps_48_kHz_in_48_kHz_out_FOA_out_object_editing_JBM_Prof_5_DISC.192

scripts/tools/Linux/networkSimulator_g192 scripts/dly_error_profiles/dly_error_profile_5.dat stv3ISM48s.wav_3_ISM_with_metadata_at_384_kbps_48_kHz_in_48_kHz_out_FOA_out_object_editing_JBM_Prof_5_DISC.192 tests/dut/param_file/enc/stv3ISM48s.wav_3_ISM_with_metadata_at_384_kbps_48_kHz_in_48_kHz_out_FOA_out_object_editing_JBM_Prof_5_DISC.netsimout tests/dut/param_file/enc/stv3ISM48s.wav_3_ISM_with_metadata_at_384_kbps_48_kHz_in_48_kHz_out_FOA_out_object_editing_JBM_Prof_5_DISC.netsimtrace 2 0

IVAS_dec -fr 20 -obj_edit NULL -no_delay_cmp -Tracefile tests/dut/param_file/dec/stv3ISM48s.wav_3_ISM_with_metadata_at_384_kbps_48_kHz_in_48_kHz_out_FOA_out_object_editing_JBM_Prof_5_DISC.dectrace -VOIP FOA 48 tests/dut/param_file/enc/stv3ISM48s.wav_3_ISM_with_metadata_at_384_kbps_48_kHz_in_48_kHz_out_FOA_out_object_editing_JBM_Prof_5_DISC.netsimout stv3ISM48s.wav_3_ISM_with_metadata_at_384_kbps_48_kHz_in_48_kHz_out_FOA_out_object_editing_JBM_Prof_5_DISC.dec.wav