Incorrect decoding of a BRR sample from the Washington theme from Time Cop

[KungFuFurby]

BRR sample ID $00 from the Washington theme from Time Cop came out in a pretty bad state when it wasn't meant to do that. Simulating the gaussian interpolation doesn't solve this issue either.

Time Cop Washington Theme Bad Sample Decoding.zip

The contents of this zip file are as following...

• out_00-noloop.wav is output from split700 as taken out of the original SPC file, which gave me the correct output.
• out_0.wav is the output from spc_decoder from the song in question. It was done from 9f6c4708150120284102f9899632d7f5f0b0638a on my previous Intel Mac running Snow Leopard.
• out0.brr is the raw BRR file. There is no loop point in the first two bytes because the sample didn't loop in the first place.
• out0BRRDecoderGauss.wav is the BRR file run through a rebuilt brr_decoder from commit 9f6c4708150120284102f9899632d7f5f0b0638a running Catalina with the -g option being used.
• out0BRRDecoderNoGauss.wav is the BRR file run through a rebuilt brr_decoder from commit 9f6c4708150120284102f9899632d7f5f0b0638a running Catalina without the -g option being used.

[nyanpasu64]

Fixed in https://github.com/nyanpasu64/BRRtools/commit/ce327ad92b517facc62b3772bc9cb3be6887760a probably.