tetrees/Plugins/Wwise/ThirdParty/include/AK/DSP/AkApplyGain.h

/*******************************************************************************
The content of this file includes portions of the AUDIOKINETIC Wwise Technology
released in source code form as part of the SDK installer package.

Commercial License Usage

Licensees holding valid commercial licenses to the AUDIOKINETIC Wwise Technology
may use this file in accordance with the end user license agreement provided 
with the software or, alternatively, in accordance with the terms contained in a
written agreement between you and Audiokinetic Inc.

Apache License Usage

Alternatively, this file may be used under the Apache License, Version 2.0 (the 
"Apache License"); you may not use this file except in compliance with the 
Apache License. You may obtain a copy of the Apache License at 
http://www.apache.org/licenses/LICENSE-2.0.

Unless required by applicable law or agreed to in writing, software distributed
under the Apache License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES
OR CONDITIONS OF ANY KIND, either express or implied. See the Apache License for
the specific language governing permissions and limitations under the License.

  Copyright (c) 2023 Audiokinetic Inc.
*******************************************************************************/

// Accumulate (+=) signal into output buffer
// To be used on mono signals, create as many instances as there are channels if need be

#ifndef _AKAPPLYGAIN_H_
#define _AKAPPLYGAIN_H_

#include <AK/SoundEngine/Common/AkTypes.h>
#include <AK/SoundEngine/Common/AkCommonDefs.h>
#include <AK/SoundEngine/Common/AkSimd.h>

#if defined (AKSIMD_V4F32_SUPPORTED) || defined (AKSIMD_V2F32_SUPPORTED)
#include <AK/Plugin/PluginServices/AkVectorValueRamp.h>
// Otherwise, it is preferrable not to use a generic implementation of a vector type.
#endif

namespace AK
{
	namespace DSP
	{
		/// Single channel, in-place interpolating gain helper (do not call directly) use ApplyGain instead.
		static inline void ApplyGainRamp(	
			AkSampleType * AK_RESTRICT io_pfBuffer, 
			AkReal32 in_fCurGain,
			AkReal32 in_fTargetGain,
			AkUInt32 in_uNumFrames )
		{
			AkSampleType * AK_RESTRICT pfBuf = (AkSampleType *) io_pfBuffer;
			const AkSampleType * const pfEnd = io_pfBuffer + in_uNumFrames;

#ifdef AKSIMD_V4F32_SUPPORTED
			const AkUInt32 uNumVecIter = in_uNumFrames/4;
			if(uNumVecIter)
			{
				CAkVectorValueRamp vGainRamp;
				AKSIMD_V4F32 vfGain = vGainRamp.Setup(in_fCurGain,in_fTargetGain,uNumVecIter*4);	
				const AkSampleType * const pfVecEnd = io_pfBuffer + uNumVecIter*4;
				while ( pfBuf < pfVecEnd )
				{
					AKSIMD_V4F32 vfIn = AKSIMD_LOAD_V4F32((AKSIMD_F32*)pfBuf);
					AKSIMD_V4F32 vfOut = AKSIMD_MUL_V4F32( vfIn, vfGain );
					AKSIMD_STORE_V4F32( (AKSIMD_F32*)pfBuf, vfOut );
					vfGain = vGainRamp.Tick();
					pfBuf+=4;
				}
			}
#elif defined (AKSIMD_V2F32_SUPPORTED)
			// Unroll 4 times x v2.
			const AkUInt32 uNumVecIter = in_uNumFrames/8;
			if(uNumVecIter)
			{
				CAkVectorValueRampV2 vGainRamp;
				AKSIMD_V2F32 vfGain = vGainRamp.Setup(in_fCurGain,in_fTargetGain,uNumVecIter*8);	
				const AkSampleType * const AK_RESTRICT pfVecEnd = io_pfBuffer + uNumVecIter*8;
				while ( pfBuf < pfVecEnd )
				{
					AKSIMD_V2F32 vfIn1 = AKSIMD_LOAD_V2F32_OFFSET( pfBuf, 0 );
					AKSIMD_V2F32 vfIn2 = AKSIMD_LOAD_V2F32_OFFSET( pfBuf, 8 );
					AKSIMD_V2F32 vfIn3 = AKSIMD_LOAD_V2F32_OFFSET( pfBuf, 16 );
					AKSIMD_V2F32 vfIn4 = AKSIMD_LOAD_V2F32_OFFSET( pfBuf, 24 );
					AKSIMD_V2F32 vfOut1 = AKSIMD_MUL_V2F32( vfIn1, vfGain );
					vfGain = vGainRamp.Tick();
					AKSIMD_V2F32 vfOut2 = AKSIMD_MUL_V2F32( vfIn2, vfGain );
					vfGain = vGainRamp.Tick();
					AKSIMD_V2F32 vfOut3 = AKSIMD_MUL_V2F32( vfIn3, vfGain );
					vfGain = vGainRamp.Tick();
					AKSIMD_V2F32 vfOut4 = AKSIMD_MUL_V2F32( vfIn4, vfGain );
					vfGain = vGainRamp.Tick();
					AKSIMD_STORE_V2F32_OFFSET( pfBuf, 0, vfOut1 );
					AKSIMD_STORE_V2F32_OFFSET( pfBuf, 8, vfOut2 );
					AKSIMD_STORE_V2F32_OFFSET( pfBuf, 16, vfOut3 );
					AKSIMD_STORE_V2F32_OFFSET( pfBuf, 24, vfOut4 );
					pfBuf+=8;
				}
			}
			/*
			const AkUInt32 uNumVecIter = in_uNumFrames/2;
			CAkVectorValueRampV2 vGainRamp;
			AKSIMD_V2F32 vfGain = vGainRamp.Setup(in_fCurGain,in_fTargetGain,uNumVecIter*2);	
			const AkSampleType * const pfVecEnd = io_pfBuffer + uNumVecIter*2;
			while ( pfBuf < pfVecEnd )
			{
				AKSIMD_V2F32 vfIn = AKSIMD_LOAD_V2F32_OFFSET( pfBuf, 0 );
				AKSIMD_V2F32 vfOut = AKSIMD_MUL_V2F32( vfIn, vfGain );
				AKSIMD_STORE_V2F32_OFFSET( pfBuf, 0, vfOut );
				vfGain = vGainRamp.Tick();
				pfBuf+=2;
			}
			*/
#endif
			if ( pfBuf < pfEnd )
			{
				const AkReal32 fInc = (in_fTargetGain - in_fCurGain) / in_uNumFrames;
				do
				{
					*pfBuf = (AkSampleType)(*pfBuf * in_fCurGain);
					in_fCurGain += fInc;
					pfBuf++;
				}while ( pfBuf < pfEnd );
			}

		}

		/// Single channel, out-of-place interpolating gain helper (do not call directly) use ApplyGain instead.
		static inline void ApplyGainRamp(	
			AkSampleType * AK_RESTRICT in_pfBufferIn, 
			AkSampleType * AK_RESTRICT out_pfBufferOut, 
			AkReal32 in_fCurGain,
			AkReal32 in_fTargetGain,
			AkUInt32 in_uNumFrames )
		{
			AkSampleType * AK_RESTRICT pfInBuf = (AkSampleType * ) in_pfBufferIn;
			AkSampleType * AK_RESTRICT pfOutBuf = (AkSampleType * ) out_pfBufferOut;
			const AkSampleType * const pfEnd = pfInBuf + in_uNumFrames;

#ifdef AKSIMD_V4F32_SUPPORTED
			const AkUInt32 uNumVecIter = in_uNumFrames/4;
			if(uNumVecIter)
			{
				CAkVectorValueRamp vGainRamp;
				AKSIMD_V4F32 vfGain = vGainRamp.Setup(in_fCurGain,in_fTargetGain,uNumVecIter*4);	
				const AkSampleType * const pfVecEnd = in_pfBufferIn + uNumVecIter*4;
				while ( pfInBuf < pfVecEnd )
				{
					AKSIMD_V4F32 vfIn = AKSIMD_LOAD_V4F32((AKSIMD_F32*)pfInBuf);
					AKSIMD_V4F32 vfOut = AKSIMD_MUL_V4F32( vfIn, vfGain );
					AKSIMD_STORE_V4F32( (AKSIMD_F32*)pfOutBuf, vfOut );
					vfGain = vGainRamp.Tick();
					pfInBuf+=4;
					pfOutBuf+=4;
				}
			}
#elif defined (AKSIMD_V2F32_SUPPORTED)
			const AkUInt32 uNumVecIter = in_uNumFrames/2;
			if(uNumVecIter)
			{
				CAkVectorValueRampV2 vGainRamp;
				f32x2 vfGain = vGainRamp.Setup(in_fCurGain,in_fTargetGain,uNumVecIter*2);	
				const AkSampleType * const pfVecEnd = in_pfBufferIn + uNumVecIter*2;
				while ( pfInBuf < pfVecEnd )
				{
					AKSIMD_V2F32 vfIn = AKSIMD_LOAD_V2F32_OFFSET( pfInBuf, 0 );
					AKSIMD_V2F32 vfOut = AKSIMD_MUL_V2F32( vfIn, vfGain );
					AKSIMD_STORE_V2F32_OFFSET( pfOutBuf, 0, vfOut );
					vfGain = vGainRamp.Tick();
					pfInBuf+=2;
					pfOutBuf+=2;
				}
			}
#endif
			if ( pfInBuf < pfEnd )
			{
				const AkReal32 fInc = (in_fTargetGain - in_fCurGain) / in_uNumFrames;
				do
				{
					*pfOutBuf++ = (AkSampleType)(*pfInBuf++ * in_fCurGain);
					in_fCurGain += fInc;
				}while ( pfInBuf < pfEnd );
			}
		}	

		/// Single channel, in-place static gain.
		static inline void ApplyGain(	
			AkSampleType * AK_RESTRICT io_pfBuffer, 
			AkReal32 in_fGain,
			AkUInt32 in_uNumFrames )
		{
			if ( in_fGain != 1.f )
			{
				AkSampleType * AK_RESTRICT pfBuf = (AkSampleType * ) io_pfBuffer;
				const AkSampleType * const pfEnd = io_pfBuffer + in_uNumFrames;

#ifdef AKSIMD_V4F32_SUPPORTED
				const AkUInt32 uNumVecIter = in_uNumFrames/4;
				if(uNumVecIter)
				{
					const AkSampleType * const pfVecEnd = io_pfBuffer + uNumVecIter*4;
					const AKSIMD_V4F32 vfGain = AKSIMD_LOAD1_V4F32( in_fGain );
					while ( pfBuf < pfVecEnd )
					{
						AKSIMD_V4F32 vfIn = AKSIMD_LOAD_V4F32((AKSIMD_F32*)pfBuf);
						AKSIMD_V4F32 vfOut = AKSIMD_MUL_V4F32( vfIn, vfGain );
						AKSIMD_STORE_V4F32( (AKSIMD_F32*)pfBuf, vfOut );
						pfBuf+=4;
					}
				}
#elif defined (AKSIMD_V2F32_SUPPORTED)
				// Unroll 4 times x 2 floats
				const AkUInt32 uNumVecIter = in_uNumFrames/8;
				if( uNumVecIter)
				{
					AKSIMD_V2F32 vfGain = __PS_FDUP( in_fGain );
					const AkSampleType * const pfVecEnd = io_pfBuffer + uNumVecIter*8;
					while ( pfBuf < pfVecEnd )
					{
						AKSIMD_V2F32 vfIn1 = AKSIMD_LOAD_V2F32_OFFSET( pfBuf, 0 );
						AKSIMD_V2F32 vfIn2 = AKSIMD_LOAD_V2F32_OFFSET( pfBuf, 8 );
						AKSIMD_V2F32 vfIn3 = AKSIMD_LOAD_V2F32_OFFSET( pfBuf, 16 );
						AKSIMD_V2F32 vfIn4 = AKSIMD_LOAD_V2F32_OFFSET( pfBuf, 24 );
						AKSIMD_V2F32 vfOut1 = AKSIMD_MUL_V2F32( vfIn1, vfGain );
						AKSIMD_V2F32 vfOut2 = AKSIMD_MUL_V2F32( vfIn2, vfGain );
						AKSIMD_V2F32 vfOut3 = AKSIMD_MUL_V2F32( vfIn3, vfGain );
						AKSIMD_V2F32 vfOut4 = AKSIMD_MUL_V2F32( vfIn4, vfGain );
						AKSIMD_STORE_V2F32_OFFSET( pfBuf, 0, vfOut1 );
						AKSIMD_STORE_V2F32_OFFSET( pfBuf, 8, vfOut2 );
						AKSIMD_STORE_V2F32_OFFSET( pfBuf, 16, vfOut3 );
						AKSIMD_STORE_V2F32_OFFSET( pfBuf, 24, vfOut4 );
						pfBuf+=8;
					}
				}
				/*
				const AkUInt32 uNumVecIter = in_uNumFrames/2;
				AKSIMD_V2F32 vfGain = __PS_FDUP( in_fGain );
				const AkSampleType * const pfVecEnd = io_pfBuffer + uNumVecIter*2;
				while ( pfBuf < pfVecEnd )
				{
					AKSIMD_V2F32 vfIn = AKSIMD_LOAD_V2F32_OFFSET( pfBuf, 0 );
					AKSIMD_V2F32 vfOut = AKSIMD_MUL_V2F32( vfIn, vfGain );
					AKSIMD_STORE_V2F32_OFFSET( pfBuf, 0, vfOut );
					pfBuf+=2;
				}
				*/
#endif
				while ( pfBuf < pfEnd )
				{
					*pfBuf = (AkSampleType)(*pfBuf * in_fGain);
					pfBuf++;
				}
			}
		}

		/// Single channel, Out-of-place static gain.
		static inline void ApplyGain(	
			AkSampleType * AK_RESTRICT in_pfBufferIn, 
			AkSampleType * AK_RESTRICT out_pfBufferOut, 
			AkReal32 in_fGain,
			AkUInt32 in_uNumFrames )
		{
			AkSampleType * AK_RESTRICT pfInBuf = (AkSampleType * ) in_pfBufferIn;
			AkSampleType * AK_RESTRICT pfOutBuf = (AkSampleType * ) out_pfBufferOut;
			const AkSampleType * const pfEnd = in_pfBufferIn + in_uNumFrames;

#ifdef AKSIMD_V4F32_SUPPORTED
			const AkUInt32 uNumVecIter = in_uNumFrames/4;
			if( uNumVecIter)
			{
				const AkSampleType * const pfVecEnd = in_pfBufferIn + uNumVecIter*4;
				const AKSIMD_V4F32 vfGain = AKSIMD_LOAD1_V4F32( in_fGain );
				while ( pfInBuf < pfVecEnd )
				{
					AKSIMD_V4F32 vfIn = AKSIMD_LOAD_V4F32((AKSIMD_F32*)pfInBuf);
					AKSIMD_V4F32 vfOut = AKSIMD_MUL_V4F32( vfIn, vfGain );
					AKSIMD_STORE_V4F32( (AKSIMD_F32*)pfOutBuf, vfOut );
					pfInBuf+=4;
					pfOutBuf+=4;
				}
			}
#elif defined (AKSIMD_V2F32_SUPPORTED)
			// Unroll 4 times x 2 floats
			const AkUInt32 uNumVecIter = in_uNumFrames/8;
			if( uNumVecIter)
			{
				const AkSampleType * const pfVecEnd = in_pfBufferIn + uNumVecIter*8;
				AKSIMD_V2F32 vfGain = __PS_FDUP( in_fGain );
				while ( pfInBuf < pfVecEnd )
				{
					AKSIMD_V2F32 vfIn1 = AKSIMD_LOAD_V2F32_OFFSET( pfInBuf, 0 );
					AKSIMD_V2F32 vfIn2 = AKSIMD_LOAD_V2F32_OFFSET( pfInBuf, 8 );
					AKSIMD_V2F32 vfIn3 = AKSIMD_LOAD_V2F32_OFFSET( pfInBuf, 16 );
					AKSIMD_V2F32 vfIn4 = AKSIMD_LOAD_V2F32_OFFSET( pfInBuf, 24 );
					pfInBuf+=8;
					AKSIMD_V2F32 vfOut1 = AKSIMD_MUL_V2F32( vfIn1, vfGain );
					AKSIMD_V2F32 vfOut2 = AKSIMD_MUL_V2F32( vfIn2, vfGain );
					AKSIMD_V2F32 vfOut3 = AKSIMD_MUL_V2F32( vfIn3, vfGain );
					AKSIMD_V2F32 vfOut4 = AKSIMD_MUL_V2F32( vfIn4, vfGain );
					AKSIMD_STORE_V2F32_OFFSET( pfOutBuf, 0, vfOut1 );
					AKSIMD_STORE_V2F32_OFFSET( pfOutBuf, 8, vfOut2 );
					AKSIMD_STORE_V2F32_OFFSET( pfOutBuf, 16, vfOut3 );
					AKSIMD_STORE_V2F32_OFFSET( pfOutBuf, 24, vfOut4 );
					pfOutBuf+=8;
				}
			}
			/*
			const AkUInt32 uNumVecIter = in_uNumFrames/2;
			const AkSampleType * const pfVecEnd = in_pfBufferIn + uNumVecIter*2;
			AKSIMD_V2F32 vfGain = __PS_FDUP( in_fGain );
			while ( pfInBuf < pfVecEnd )
			{
				AKSIMD_V2F32 vfIn = AKSIMD_LOAD_V2F32_OFFSET( pfInBuf, 0 );
				AKSIMD_V2F32 vfOut = AKSIMD_MUL_V2F32( vfIn, vfGain );
				AKSIMD_STORE_V2F32_OFFSET( pfOutBuf, 0, vfOut );
				pfInBuf+=2;
				pfOutBuf+=2;
			}
			*/
#endif
			while ( pfInBuf < pfEnd )
			{
				*pfOutBuf++ = (AkSampleType)(*pfInBuf++ * in_fGain);
			}
		}


		/// Single channel, In-place (possibly interpolating) gain.
		static inline void ApplyGain(	
			AkSampleType * AK_RESTRICT io_pfBuffer, 
			AkReal32 in_fCurGain,
			AkReal32 in_fTargetGain,
			AkUInt32 in_uNumFrames )
		{

			if ( in_fTargetGain == in_fCurGain )
				ApplyGain(io_pfBuffer, in_fCurGain, in_uNumFrames );
			else
				ApplyGainRamp( io_pfBuffer, in_fCurGain, in_fTargetGain, in_uNumFrames );
		}

		/// Single channel, Out-of-place (possibly interpolating) gain.
		static inline void ApplyGain(	
			AkSampleType * AK_RESTRICT in_pfBufferIn, 
			AkSampleType * AK_RESTRICT out_pfBufferOut, 
			AkReal32 in_fCurGain,
			AkReal32 in_fTargetGain,
			AkUInt32 in_uNumFrames )
		{
			if ( in_fTargetGain == in_fCurGain )
				ApplyGain(in_pfBufferIn, out_pfBufferOut, in_fCurGain, in_uNumFrames );
			else
				ApplyGainRamp( in_pfBufferIn, out_pfBufferOut, in_fCurGain, in_fTargetGain, in_uNumFrames );
		}

		/// Multi-channel in-place (possibly interpolating) gain.
		static inline void ApplyGain( 
			AkAudioBuffer * io_pBuffer,
			AkReal32 in_fCurGain,
			AkReal32 in_fTargetGain,
			bool in_bProcessLFE = true )
		{
			AkUInt32 uNumChannels = io_pBuffer->NumChannels();
			if ( !in_bProcessLFE && io_pBuffer->HasLFE() )
				uNumChannels--;
			const AkUInt32 uNumFrames = io_pBuffer->uValidFrames;
			if ( in_fTargetGain == in_fCurGain )
			{
				// No need for interpolation
				for ( AkUInt32 i = 0; i < uNumChannels; i++ )
				{
					AkSampleType * pfChan = io_pBuffer->GetChannel( i );
					ApplyGain(pfChan, in_fCurGain, uNumFrames );
				}
			}
			else
			{
				// Interpolate gains toward target
				for ( AkUInt32 i = 0; i < uNumChannels; i++ )
				{
					AkSampleType * pfChan = io_pBuffer->GetChannel( i );
					ApplyGainRamp(pfChan, in_fCurGain, in_fTargetGain, uNumFrames );
				}
			}
		}

		/// Single-channel LFE in-place (possibly interpolating) gain.
		static inline void ApplyGainLFE( 
			AkAudioBuffer * io_pBuffer,
			AkReal32 in_fCurGain,
			AkReal32 in_fTargetGain )
		{
			if( io_pBuffer->HasLFE() )
			{
				AkUInt32 uLFEChannelIdx = io_pBuffer->NumChannels()-1;
				const AkUInt32 uNumFrames = io_pBuffer->uValidFrames;
				AkSampleType * pfChan = io_pBuffer->GetChannel( uLFEChannelIdx );
				if ( in_fTargetGain == in_fCurGain )
				{
					// No need for interpolation
					ApplyGain(pfChan, in_fCurGain, uNumFrames );
				}
				else
				{
					// Interpolate gains toward target
					ApplyGainRamp(pfChan, in_fCurGain, in_fTargetGain, uNumFrames );
				}
			}
		}

		/// Multi-channel out-of-place (possibly interpolating) gain.
		static inline void ApplyGain( 
			AkAudioBuffer * in_pBuffer,
			AkAudioBuffer * out_pBuffer,
			AkReal32 in_fCurGain,
			AkReal32 in_fTargetGain,
			bool in_bProcessLFE = true )
		{
			AKASSERT( in_pBuffer->NumChannels() == out_pBuffer->NumChannels() );
			AkUInt32 uNumChannels = in_pBuffer->NumChannels();
			if ( !in_bProcessLFE && in_pBuffer->HasLFE() )
				uNumChannels--;
			const AkUInt32 uNumFrames = AkMin( in_pBuffer->uValidFrames, out_pBuffer->MaxFrames() );
			if ( in_fTargetGain == in_fCurGain )
			{
				// No need for interpolation
				for ( AkUInt32 i = 0; i < uNumChannels; i++ )
				{
					AkSampleType * pfInChan = in_pBuffer->GetChannel( i );
					AkSampleType * pfOutChan = out_pBuffer->GetChannel( i );
					ApplyGain(pfInChan, pfOutChan, in_fCurGain, uNumFrames );
				}
			}
			else
			{
				// Interpolate gains toward target
				for ( AkUInt32 i = 0; i < uNumChannels; i++ )
				{
					AkSampleType * pfInChan = in_pBuffer->GetChannel( i );
					AkSampleType * pfOutChan = out_pBuffer->GetChannel( i );
					ApplyGainRamp( pfInChan, pfOutChan, in_fCurGain, in_fTargetGain, uNumFrames );
				}
			}
		}

	} // namespace DSP
} // namespace AK

#endif // _AKAPPLYGAIN_H_