kraken_engine/kraken
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Changed the buffer constants to decrease the number of drop outs (still needs some work)

Browse Source 
--HG--
branch : nfb
This commit is contained in:
Peter Courtemanche
2014-01-14 15:51:32 -08:00
parent edbae68dc5
commit ca1af8dd3e
1 changed files with 15 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

21
KREngine/kraken/KRAudioManager.h
View File

@@ -40,18 +40,27 @@
#include "KRMat4.h"
#include "KRAudioSource.h"
const int KRENGINE_AUDIO_MAX_POOL_SIZE = 32;
const int KRENGINE_AUDIO_MAX_BUFFER_SIZE = 64*1024; // this is the buffer for our decoded audio (not the source file data)
const int KRENGINE_AUDIO_MAX_POOL_SIZE = 40; //32;
// for Circa we play a maximum of 7 mono audio streams at once + cross fading with ambient
// so we could safely say a maximum of 10 streams, which would be 33 buffers
const int KRENGINE_AUDIO_MAX_BUFFER_SIZE = 5120; // in bytes
// this is the buffer for our decoded audio (not the source file data)
// it should be greater then 1152 samples (the size of an mp3 frame in samples)
// so it should be greater then 2304 bytes and also a multiple of 128 samples (to make
// the data flow efficient) but it shouldn't be too large or it will cause
// the render loop to stall out decoding large chunks of mp3 data.
// 2560 bytes would be the smallest size for mono sources, and 5120 would be smallest for stereo.
const int KRENGINE_AUDIO_BUFFERS_PER_SOURCE = 3;
const int KRENGINE_AUDIO_BLOCK_LOG2N = 7; // 2 ^ KRENGINE_AUDIO_BLOCK_LOG2N = KRENGINE_AUDIO_BLOCK_LENGTH
// 7 is 128, 8 -> 256, 9 -> 512, 10 -> 1024 (the size of the hardware (AUgraph) framebuffer)
// NOTE: the hrtf code use magic numbers everywhere and is hardcoded to 128 samples per frame
// 7 is 128 .. NOTE: the hrtf code uses magic numbers everywhere and is hardcoded to 128 samples per frame
const int KRENGINE_AUDIO_BLOCK_LENGTH = 1 << KRENGINE_AUDIO_BLOCK_LOG2N;
// Length of one block to process. Determines the latency of the audio system and sets size for FFT's used in HRTF convolution
// the AUGraph works in 1024 sample chunks. If we put a value of less then 1024 in here then we are making mutliple calls to our render functions without any
// improvement in latency and our audio render perfomance goes down significantly
// the AUGraph works in 1024 sample chunks. At 128 we are making 8 consecutive calls to the renderBlock method for each
// render initiated by the AUGraph.
const int KRENGINE_REVERB_MAX_FFT_LOG2 = 15;
const int KRENGINE_REVERB_WORKSPACE_SIZE = 1 << KRENGINE_REVERB_MAX_FFT_LOG2;