Signal chain in float32_t

[probonopd]

Closes #71

[dcoredump]

Now I got upstream merged into my fork and switched the complete audio signal path into type float32_t. MiniDexeds output seems to be louder now (without clipping). The reverb still has a noisy tail, but I think this is a problem of the PWM output. I hope I can build a setup with a PCM5102A soon.

[probonopd]

Thanks @dcoredump. Significant improvement. Tested good with HDMI on RPi 2.

[rsta2]

CXXFLAGS += -DARM_MATH_NEON -DHAVE_NEON in Synth_Dexed.mk has no effect. Use DEFINE += -DARM_MATH_NEON -DHAVE_NEON instead to define these macros.

[dcoredump]

CXXFLAGS += -DARM_MATH_NEON -DHAVE_NEON in Synth_Dexed.mk has no effect. Use DEFINE += -DARM_MATH_NEON -DHAVE_NEON instead to define these macros.

Thanks @rsta2 .

You mentioned this earlier and I forgot to fix this in my fork.

[rsta2]

I did some "code review" to learn about the changes. I think ProcessSound() without ARM_ALLOW_MULTI_CORE for the RPi 1 produces Stereo audio now, but is initialized for Mono here. That does not match. It's good that the output is much louder now. For what is the file src/minidexed.cpp.O?