[Performous-commit] Lasse Karkkainen : Workaround the audio analysis bug ( the real fix would go to ffmpeg.*)

[Lasse Kärkkäi. <tr...@us...>]

Module: editor
Branch: master
Commit: 60164a8f8b9be7be3cb90a8e53518272dbf1c595

Author: Lasse Karkkainen <tro...@tr...>
Date:   Tue Jan 25 17:04:24 2011 +0100

Workaround the audio analysis bug (the real fix would go to ffmpeg.*)

---

 pitchvis.cc |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/pitchvis.cc b/pitchvis.cc
index d3b1678..64ecae5 100644
--- a/pitchvis.cc
+++ b/pitchvis.cc
@@ -36,15 +36,16 @@ void PitchVis::run()
 			}
 			msleep(1000); // Wait some more
 			setWidth(mpeg.duration() * rate / step); // Estimation
-
 			curX = 0;
 			for (std::vector<float> data(step*2); mpeg.audioQueue(&*data.begin(), &*data.end(), curX * step * 2); ++curX) {
+				msleep(2);
 				// Mix stereo into mono
 				for (unsigned i = 0; i < step; ++i) data[i] = 0.5 * (data[2*i] + data[2*i + 1]);
 				// Process
 				analyzer.input(&data[0], &data[step]);
 				analyzer.process();
 				if (cancelled) return;
+				std::fill(data.begin(), data.end(), 0.0f);
 			}
 		}
 		// Draw, TODO: Draw somewhere else (on demand) without a huge bitmap