You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Mik made this very useful patch back in January, i put it here on the Wishlist for reference.
Especially for very dynamic songs the RMS is way better then what we use today.
<snip>
Hello.
Waveform view on mixxx looks very spiky. This results from peak detection
algorithm that extract waveform envelop. Since in my eyes rms averaging is the
better, if programmed it into mixxx. On the links below one can compare peak
detection and rms averaging on a techno track and a rock song. It doesn't look
that much better like I had expected but when playing sound files even the
small spikes are very helpful by getting a feeling of the beats and rhythm
hidden behind the waveform.
Since rms averaging requiers a lot cpu time performance is a critical issue. A
tricky alogrithm can reduce calculation afford with minor loss results. In
fact my variant is as fast as peak detection.
The attachment contains a patch made from revision 2293 with the changes I've
applied to implement rms averaging. The code is made with limit checking thus
at least critical bugs which make mixxx instable are very unlikely.
Also there is an open issue. I'm not a daily C++ programmer but if
WaveformAnalyser object is destroy QVector<float> is freed as well.
In hope I could help
mik
</snip>
The text was updated successfully, but these errors were encountered:
Reported by: esbrandt
Date: 2010-03-04T18:43:02Z
Status: Fix Released
Importance: Wishlist
Launchpad Issue: lp532090
Tags: waveform
Attachments: rms.patch
Mik made this very useful patch back in January, i put it here on the Wishlist for reference.
Especially for very dynamic songs the RMS is way better then what we use today.
Waveform view on mixxx looks very spiky. This results from peak detection
algorithm that extract waveform envelop. Since in my eyes rms averaging is the
better, if programmed it into mixxx. On the links below one can compare peak
detection and rms averaging on a techno track and a rock song. It doesn't look
that much better like I had expected but when playing sound files even the
small spikes are very helpful by getting a feeling of the beats and rhythm
hidden behind the waveform.
http://img641.imageshack.us/i/mixxxpeak.png/
http://img52.imageshack.us/i/mixxxrms.png/
Since rms averaging requiers a lot cpu time performance is a critical issue. A
tricky alogrithm can reduce calculation afford with minor loss results. In
fact my variant is as fast as peak detection.
The attachment contains a patch made from revision 2293 with the changes I've
applied to implement rms averaging. The code is made with limit checking thus
at least critical bugs which make mixxx instable are very unlikely.
In hope I could help
The text was updated successfully, but these errors were encountered: