Mainly noise MASA selection test signal has clear spectral differences with BASOP encoder

Basic info

• Float reference:
  • Encoder (float): db364042
  • Decoder (float): db364042
• Fixed point:
  • Encoder (fixed): f1458362
  • Decoder (fixed): f1458362

Bug description

There seems to be a clear spectral or level difference in MASA selection test sample containing mainly noise signal. As this is EXT output, the difference is in the transports. Based on listening, there are at least two clear parts with difference.

• At 4-5 s, the BASOP one seems to contain "something more"
• At 7-9 s, the noise envelope is different
• At 16-18 s, the BASOP one is missing something (this shows in plot as well)

Here is time domain plot of full float vs. BASOP enc to float dec.

Ways to reproduce

Box folder: ...\Box_EXTERNAL_IVAS_BASOP_VERIFICATION\issues\issue-1775

Inputs are in Box

IVAS_cod -masa 2 BASOP_Selection_MUSHRA-7ad_s04.met -max_band wb -dtx 32000 48 BASOP_Selection_MUSHRA-7ad_s04.wav test_1000.bs
IVAS_dec EXT 48 test_1000.bs 1000_BASOP_Selection_MUSHRA-7ad_s04.wav_32k_48-48___EXT.wav