Fix the game using too many Wii U voices

Now it just allocates channels on-demand, rather than ahead-of-time.

src/Backends/Audio/WiiU.cpp

@@ -18,7 +18,12 @@
 struct AudioBackend_Sound
 {
 	unsigned char *samples;
+	size_t length;
 	AXVoice *voice;
+	unsigned int frequency;
+	unsigned short volume;
+	unsigned short pan_l;
+	unsigned short pan_r;
 };
 
 static double MillibelToScale(long volume)
@@ -59,55 +64,23 @@ AudioBackend_Sound* AudioBackend_CreateSound(unsigned int frequency, const unsig
 
 		if (samples_copy != NULL)
 		{
+			// Convert to signed
 			for (size_t i = 0; i < length; ++i)
 				samples_copy[i] = samples[i] - 0x80;
 
 			DCStoreRange(samples_copy, length);
 
-			AXVoice *voice = AXAcquireVoice(31, NULL, NULL);
-
-			if (voice != NULL)
-			{
-				AXVoiceOffsets offs;
-				AXVoiceVeData vol = {
-					.volume = 0x8000,
-				};
-
-				AXVoiceBegin(voice);
-
-				AXSetVoiceType(voice, 0);
-				AXSetVoiceVe(voice, &vol);
-
-				static AXVoiceDeviceMixData mix_data[1][6];
-				mix_data[0][0].bus[0].volume = 0x8000;
-				mix_data[0][1].bus[0].volume = 0x8000;
-
-				AXSetVoiceDeviceMix(voice, AX_DEVICE_TYPE_DRC, 0, mix_data[0]);
-				AXSetVoiceDeviceMix(voice, AX_DEVICE_TYPE_TV, 0, mix_data[0]);
-
-				float srcratio = (float)frequency / (float)AXGetInputSamplesPerSec();
-				AXSetVoiceSrcRatio(voice, srcratio);
-				AXSetVoiceSrcType(voice, AX_VOICE_SRC_TYPE_LINEAR);
-
-				offs.dataType = AX_VOICE_FORMAT_LPCM8;
-				offs.endOffset = length;
-				offs.loopingEnabled = AX_VOICE_LOOP_DISABLED;
-				offs.loopOffset = 0;
-				offs.currentOffset = 0;
-				offs.data = samples_copy;
-				AXSetVoiceOffsets(voice, &offs);
-
-				AXVoiceEnd(voice);
-
 			sound->samples = samples_copy;
-				sound->voice = voice;
+			sound->length = length;
+			sound->voice = NULL;
+			sound->frequency = frequency;
+			sound->volume = 0x8000;
+			sound->pan_l = 0x8000;
+			sound->pan_r = 0x8000;
 
 			return sound;
 		}
 
-			free(samples_copy);
-		}
-
 		free(sound);
 	}
 
@@ -116,41 +89,100 @@ AudioBackend_Sound* AudioBackend_CreateSound(unsigned int frequency, const unsig
 
 void AudioBackend_DestroySound(AudioBackend_Sound *sound)
 {
-	AXFreeVoice(sound->voice);
+	AudioBackend_StopSound(sound);
+
 	free(sound->samples);
 	free(sound);
 }
 
 void AudioBackend_PlaySound(AudioBackend_Sound *sound, bool looping)
 {
+	if (sound->voice == NULL)
+	{
+		AXVoice *voice = AXAcquireVoice(31, NULL, NULL);
+
+		if (voice != NULL)
+		{
+			AXVoiceOffsets offs;
+			AXVoiceVeData vol = {
+				.volume = sound->volume,
+			};
+
+			AXVoiceBegin(voice);
+
+			AXSetVoiceType(voice, 0);
+			AXSetVoiceVe(voice, &vol);
+
+			static AXVoiceDeviceMixData mix_data[1][6];
+			mix_data[0][0].bus[0].volume = sound->pan_l;
+			mix_data[0][1].bus[0].volume = sound->pan_r;
+
+			AXSetVoiceDeviceMix(voice, AX_DEVICE_TYPE_DRC, 0, mix_data[0]);
+			AXSetVoiceDeviceMix(voice, AX_DEVICE_TYPE_TV, 0, mix_data[0]);
+
+			float srcratio = (float)sound->frequency / (float)AXGetInputSamplesPerSec();
+			AXSetVoiceSrcRatio(voice, srcratio);
+			AXSetVoiceSrcType(voice, AX_VOICE_SRC_TYPE_LINEAR);
+
+			offs.dataType = AX_VOICE_FORMAT_LPCM8;
+			offs.endOffset = sound->length;
+			offs.loopingEnabled = AX_VOICE_LOOP_DISABLED;
+			offs.loopOffset = 0;
+			offs.currentOffset = 0;
+			offs.data = sound->samples;
+			AXSetVoiceOffsets(voice, &offs);
+
+			AXVoiceEnd(voice);
+
+			sound->voice = voice;
+		}
+	}
+
+	if (sound->voice != NULL)
+	{
 		AXSetVoiceLoop(sound->voice, looping ? AX_VOICE_LOOP_ENABLED : AX_VOICE_LOOP_DISABLED);
 		AXSetVoiceState(sound->voice, AX_VOICE_STATE_PLAYING);
+	}
 }
 
 void AudioBackend_StopSound(AudioBackend_Sound *sound)
 {
-	AXSetVoiceState(sound->voice, AX_VOICE_STATE_STOPPED);
+	if (sound->voice != NULL)
-	AXSetVoiceCurrentOffset(sound->voice, 0);
+	{
+		AXFreeVoice(sound->voice);
+		sound->voice = NULL;
+	}
 }
 
 void AudioBackend_RewindSound(AudioBackend_Sound *sound)
 {
+	if (sound->voice != NULL)
 		AXSetVoiceCurrentOffset(sound->voice, 0);
 }
 
 void AudioBackend_SetSoundFrequency(AudioBackend_Sound *sound, unsigned int frequency)
 {
+	sound->frequency = frequency;
+
+	if (sound->voice != NULL)
+	{
 		float srcratio = (float)frequency / (float)AXGetInputSamplesPerSec();
 		AXSetVoiceSrcRatio(sound->voice, srcratio);
+	}
 }
 
 void AudioBackend_SetSoundVolume(AudioBackend_Sound *sound, long volume)
 {
+	sound->volume = (unsigned short)(0x8000 * MillibelToScale(volume));
+
+	if (sound->voice != NULL)
+	{
 		AXVoiceVeData vol = {
-		.volume = (unsigned short)(0x8000 * MillibelToScale(volume)),
+			.volume = sound->volume,
 		};
 
 		AXSetVoiceVe(sound->voice, &vol);
+	}
 }
 
 void AudioBackend_SetSoundPan(AudioBackend_Sound *sound, long pan)
@@ -159,8 +191,11 @@ void AudioBackend_SetSoundPan(AudioBackend_Sound *sound, long pan)
 	mix_data[0][0].bus[0].volume = (unsigned short)(0x8000 * MillibelToScale(-pan));
 	mix_data[0][1].bus[0].volume = (unsigned short)(0x8000 * MillibelToScale(pan));
 
+	if (sound->voice != NULL)
+	{
 		AXSetVoiceDeviceMix(sound->voice, AX_DEVICE_TYPE_DRC, 0, mix_data[0]);
 		AXSetVoiceDeviceMix(sound->voice, AX_DEVICE_TYPE_TV, 0, mix_data[0]);
+	}
 }
 
 void AudioBackend_SetOrganyaCallback(void (*callback)(void), unsigned int milliseconds)