Switch to floating-point sound internally.

This allows us to change the volume of sounds in filters ("volume",
"fade" properties) without quantizing the audio prematurely.  Previously,
we had a hack to apply volume changes as a mixing-time step.  This
can be done at any point in the filter chain now.

This also means we only need to worry about clipping at the very
last mixing step, instead of at every place we change sound.

Dithering for 16-bit audio and more intelligent gain control
is possible, as we have full-resolution, unclipped sound at the
mixing phase.

We support high resolution source sounds; both Vorbis and MAD
supply more than 16 bits.

Several filters, such as resampling, are easier to implement as
floats; these no longer need to convert back and forth.

Negatives:
 - More memory use.  The main case of this is in RageSoundReader_Preload,
which will be fixed: we can preload in 16-bit without losing most of the
above.
 - Some extra overhead for accessing more memory.

stepmania/src/RageSoundReader_MP3.cpp

@@ -204,7 +204,8 @@ struct madlib_t
 		bitrate = 0;
 	}
 
-	uint8_t inbuf[16384], outbuf[8192];
+	uint8_t inbuf[16384];
+	float outbuf[8192];
 	int outpos;
 	unsigned outleft;
 
@@ -268,7 +269,7 @@ struct madlib_t
 
 
 
-static signed int scale(mad_fixed_t sample)
+static float scale(mad_fixed_t sample)
 {
 	/* round */
 	sample += (1L << (MAD_F_FRACBITS - 16));
@@ -279,8 +280,7 @@ static signed int scale(mad_fixed_t sample)
 	else if (sample < -MAD_F_ONE)
 		sample = -MAD_F_ONE;
 
-	/* quantize */
-	return sample >> (MAD_F_FRACBITS + 1 - 16);
+	return (double) sample / (1<<MAD_F_FRACBITS);
 }
 
 static int get_this_frame_byte( const madlib_t *mad )
@@ -542,9 +542,9 @@ void RageSoundReader_MP3::synth_output()
 	{
 		for(int chan = 0; chan < this->Channels; ++chan)
 		{
-			short Sample = (short) scale(mad->Synth.pcm.samples[chan][i]);
-			*((short *) (mad->outbuf + mad->outleft)) = Sample;
-			mad->outleft += 2;
+			float Sample = scale(mad->Synth.pcm.samples[chan][i]);
+			mad->outbuf[mad->outleft] = Sample;
+			++mad->outleft;
 		}
 	}
 }
@@ -743,7 +743,7 @@ static void mono_to_stereo( char *dst, const char *src, unsigned len )
 	}
 }
 
-int RageSoundReader_MP3::Read( int16_t *buf, int iFrames )
+int RageSoundReader_MP3::Read( float *buf, int iFrames )
 {
 	int iFramesWritten = 0;
 
@@ -751,17 +751,17 @@ int RageSoundReader_MP3::Read( int16_t *buf, int iFrames )
 	{
 		if( mad->outleft > 0 )
 		{
-			int iFramesToCopy = min( iFrames, int(mad->outleft / (sizeof(int16_t) * GetNumChannels())) );
+			int iFramesToCopy = min( iFrames, int(mad->outleft / GetNumChannels()) );
 			const int iSamplesToCopy = iFramesToCopy * GetNumChannels();
-			const int iBytesToCopy = iSamplesToCopy * sizeof(int16_t);
+			const int iBytesToCopy = iSamplesToCopy * sizeof(float);
 
 			memcpy( buf, mad->outbuf + mad->outpos, iBytesToCopy );
 
 			buf += iSamplesToCopy;
 			iFrames -= iFramesToCopy;
 			iFramesWritten += iFramesToCopy;
-			mad->outpos += iBytesToCopy;
-			mad->outleft -= iBytesToCopy;
+			mad->outpos += iSamplesToCopy;
+			mad->outleft -= iSamplesToCopy;
 			continue;
 		}
 
@@ -936,10 +936,9 @@ int RageSoundReader_MP3::SetPosition_hard( int iFrame )
 
 			int samples = mad_timer_count( skip, (mad_units) SampleRate );
 
-			/* Skip 'samples' samples; each sample is 2 * channels bytes, since
-			 * we're currently always using AUDIO_S16SYS. */
-			mad->outpos = samples * 2 * this->Channels;
-			mad->outleft -= samples * 2 * this->Channels;
+			/* Skip 'samples' samples. */
+			mad->outpos = samples * this->Channels;
+			mad->outleft -= samples * this->Channels;
 			return 1;
 		}
 
@@ -1052,7 +1051,7 @@ bool RageSoundReader_MP3::SetProperty( const RString &sProperty, float fValue )
 int RageSoundReader_MP3::GetNextSourceFrame() const
 {
 	int iFrame = mad_timer_count( mad->Timer, mad_units(mad->Frame.header.samplerate) );
-	iFrame += mad->outpos / (sizeof(int16_t) * this->Channels);
+	iFrame += mad->outpos / this->Channels;
 	return iFrame;
 }