I've just checked and Ganryu doesn't use any encryption on V roms (actually the PROG rom is plain standard, without encryption on P or V roms), also, the CHA board uses the CMC 042 that doesn't encrypt the audio program, so those clicks are not caused by wrong decryption, and so neosd is running with the exact same data than MAME. Keep in mind that MAME has a software emulation of the YM2610 sound chip, and that it might not be 100% accurate to the actual hardware device, so I'd say the clicks aren't caused by neosd (or the bootleg) but they are either a bad dump, or they are also present in the actual cart. Can anyone with an original Ganryu cart confirm? I think they are quite rare though.
There could be another explanation though, and that's something I'd need to check. That the game is addressing a sample outside of the 4MB rom space that is actually used. Tomorrow I'll check erasing the rest of the V rom data, and also replicating the V rom 4 times to cover the entire V rom flash space to check if it changes.
EDIT: I couldn't wait and I tried ganryu un an emulator. The intro is just 2 long ADPCM-A samples, a first one that plays the "slow" part of the intro, looped 3 times at the beginning, and then another sample, looped other 3 times, during the "faster music" part. This second sample is the one with issues. There are not 3 clicks, but just it's just the same one repated 3 times because of the looping. All data is ok, well in range, so, there is probably just a single wrong byte.
This means that the cause is one of the 2 I mentioned: either a single byte bad dump (or probably more), or the original game just sounds that way, and the encoder the used didn't produce a proper ADPCM value, probably in some edge case. Ganryu doesn't seem a very polished game, even the change from the slow to fast music is abrupt and not properly synchronized, so that issue being in the original game looks "normal" .
Maybe in the hardware the decoder overflows, causing a click, and in mame it's clamped to sane values.