Extending supported sample rate from 3.0KHz to 192.0KHz for OfflineAudioContext.

Source/platform/audio/HRTFPanner.cpp

 /*
  * Copyright (C) 2010, Google Inc. All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions
  * are met:
  * 1.  Redistributions of source code must retain the above copyright
  *    notice, this list of conditions and the following disclaimer.
  * 2.  Redistributions in binary form must reproduce the above copyright
  *    notice, this list of conditions and the following disclaimer in the
@@ skipping 59 matching lines @@
 
 HRTFPanner::~HRTFPanner()
 {
 }
 
 size_t HRTFPanner::fftSizeForSampleRate(float sampleRate)
 {
     // The HRTF impulse responses (loaded as audio resources) are 512 sample-frames @44.1KHz.
     // Currently, we truncate the impulse responses to half this size, but an FFT-size of twice impulse response size is needed (for convolution).
     // So for sample rates around 44.1KHz an FFT size of 512 is good. We double the FFT-size only for sample rates at least double this.
-    ASSERT(sampleRate >= 44100 && sampleRate <= 96000.0);
-    return (sampleRate < 88200.0) ? 512 : 1024;
+    ASSERT(sampleRate >= 3000 && sampleRate <= 192000);
+
+    if (sampleRate < 5512.5)

[Raymond Toy, 2014/09/05 21:01:43]

Actually, I was just looking. I think we can do better than this.

In HRTFElevation.cpp, we sample rate convert the responses to the new frequency.
According to the comment in line 77, the response are 512 samples at 44100, but
it seems we only use the first 256. When we resample the response, we will get a
new length (as done in AudioBus::createBySampleRateConverting). We can use this
length to set the fft size by rounding the length up to the nearest power of 2
and then multiply by 2.

So for 44100, the length is 256, so the FFT size is 512. For 88200, the length
would be 512, so the FFT size is 1024. Finally, for 44101, the length is
truncate(256.0058) = 256, so the FFT size is 512.

What do you think?

[KhNo, 2014/09/05 21:29:16]

    int impulseLength = 256;
    float impulseSampleRate = 44100;
    double ratio = sampleRate / impulseSampleRate;

    return floor(impulseLength * ratio) * 2;

Is it similar with your proposal? Am I right?
It is good idea for explain where magic number is from.

One question, fftSize only can be a 2^n.
Do you have idea how we have to change the last line?
return floor(impulseLength * ratio) * 2;

+        return 32;
+    if (sampleRate < 11025)
+        return 64;
+    if (sampleRate < 22050)
+        return 128;
+    if (sampleRate < 44100)
+        return 256;
+    if (sampleRate < 88200)
+        return 512;
+    if (sampleRate < 176400)
+        return 1024;
+
+    return 2048;
 }
 
 void HRTFPanner::reset()
 {
     m_convolverL1.reset();
     m_convolverR1.reset();
     m_convolverL2.reset();
     m_convolverR2.reset();
     m_delayLineL.reset();
     m_delayLineR.reset();
@@ skipping 216 matching lines @@
 
 void HRTFPanner::trace(Visitor* visitor)
 {
     visitor->trace(m_databaseLoader);
     Panner::trace(visitor);
 }
 
 } // namespace blink
 
 #endif // ENABLE(WEB_AUDIO)