# Basic info <!--- Add commit SHA used to reproduce --> - Float reference: - Encoder (float): d3f01fcfb24df875b610c5c83cb1739811073deb - Decoder (float): d3f01fcfb24df875b610c5c83cb1739811073deb - Fixed point: - Decoder (fixed): 3b2f07138c61dcf997bbf4165d0882f794b2995f # Bug description There seems to be some kind of systematic difference in the metadata output in MASA EXT output mode. Namely, the directToTotal-ratio values and diffuseToTotal-ratio values in the final `extOutMeta`-structure may have difference of one quite often. With low directToTotal-ratio values, this is not so audible but this does generate unnecessary difference in metadata output. I am not sure how the difference come to be but it is probably limited to function pair `create_masa_ext_out_meta_fx` and `create_masa_ext_out_meta`. It might be this part ```C IF( EQ_32( hQMetaData->q_direction[dir].band_data[b_old].energy_ratio_fx[sf], ONE_IN_Q30 ) ) { hQMetaData->q_direction[dir].band_data[b_old].energy_ratio_fx[sf]--; } UWord8 tmp = (UWord8) L_shr( hQMetaData->q_direction[dir].band_data[b_old].energy_ratio_fx[sf], Q30 - 8 ); // Q8 move16(); extOutMeta->directToTotalRatio[dir][sf][b_new] = tmp; // Q8 move16(); ``` When considering how much this affects, the result is that significantly more frames are affected than are not. # Ways to reproduce Happens with many operating points but, for example, this is easy for testing as the difference is present already in second frame. <!-- Commandline or script --> ```bash IVAS_cod -masa 1 ltv48_MASA1TC.met -max_band wb 13200 32 ltv32_MASA1TC.wav test_0103.bs IVAS_dec EXT 48 test_0103.bs 0103_ltv32_MASA1TC.wav_13k2_32-48___EXT.wav ``` <!--- Below are labels that will be added but are not shown in description. This is a template to help fill them. Add further information to the first row and remove and add labels as necessary. -->