jlinkwitzriley phase issue
-
jlinkwitzriley nodes are causing big phase issues here (3 band version). It gives delay bigger than 50 samples. I am using these with their AP modules too.
Does anyone have a tip or a working snippet on this?
-
@Steve-Mohican are you using the template? And 50 samples seems unrealistic there is nothing in there that could hold this amount of data, so make sure it‘s not something else.
-
I built myself and here is the simple snippet below. As you see in the below, there is a big phasing problem on the band split points.
There is a frame container because this will be used in a snappy multi band compressor.
HiseSnippet 1450.3oc6Y0saaaCElJ1z4mlMz1sArK8ko.cFRx1wwXWrVmDuYr4Ti5rfcWAiDcLqkHUknah6vte.6UnWtmgc8dF1SxdC1NTTNVxww0wIo8Fq.DnygmC424WRK1IT3PihDgHiMOdT.EYrMt6Htr+98ILNp0AHiOG2lDIogE0rZLJfDEQcQFF49dECiMyihe92uqAwivcnSXgPmHXNzeh4yjS314Y+HyyqIwkdLyOkzUdVKGAeegmXHfmbXST.wY.4L5QDkXqgQFENzkIEgckDIMBjogvcT29hy4Z4OgEwN0ipHrPcgIRyFseelmamw1ZDBYjuyDKOm1x+RbalK6R9S7.OLdfhSzHsOvXs4AIqa.jLRAo7ZH8HbWmPVfbxHJ77.bKNDP5Q.WcZnnkEY7238Ef.bYIex.ZyPf3RE1YWSymVD92S91dC4NRlfWTvORHoufuyS15W2Zys9ssJN8P85MywTKSnvyiFNygUQ2v4o3N7g9mRCeZw2R7FRuTPv7y5SKrX9TGsUmRPAuEmIeQ.MgtovyU4qTue0H.JwsAu8ysNfHIpfRBOPt.ZnjofiwAz2BY05Pzl3CnQCjh.Hu9JwOHyQ3NziHylNopaRF.7GYhgp.EOhIGkttZ5br7Wu+vLk+3CXsKHbeDtCS5ze13csYfWvqcei2jJ1OCeXudTG4DvlG27Wla4Y9EKUZBTVzR0sRJUg0WCjFcmYw4uimTHDDRCHgziEc7Hi1Ih3G3QeIfymV7TOgyftr2QuZ0SfdoanjXGm9DNm5EsLEYEtiZbgdoXnjwOqMQFxt.YfOZneWnEtCc+DzA7LVSUeooMUzp7jtTtaLw+AOICZonMRFzZ7foJMOhJOWDNHNLj7NxXcsO+zHzy87DmuuvOfkjsB99XdcDdiB5K3LGEKsDiQ3y8ECgIOAl+.I5XBySkd2cXDz7v8E7tfvwasYj6DZXT7DuN1rD7GjtejvEFB2j3.NzQcHx9p5FU+HHwjFVxIUBJfwz4EJUGukm1q.PyWYpwa3kKUiGcdmhBp1VSMaaiaEchZTGhG5DU7UAZjZN09mXbUPIJF2HJ.8AP3DfoB0og113J1l0pZW1bupHUVRGRHrqLrqRjNCIKpmBAX8AI5oTw9UwI2YgxWmBJYj55Qjh9CunqiiB7fCebsFtd3kYYxkYYzNvk1+tXKyCwu1iwGbNS9tPHgbT106qvu1MJnzThbSVSUUgG8hw680j4EGfu7UUZG2kdgJO9arPGBMVbg9jwxqlATprBfwkT.7ay3IInEv1ppl1jKRXrIv.d.dcGPOWaRpiiYVx1tdcyJU201tbk8pAYd5i+zLj9lgP43HzyGJE9P+qw6ZL6UDiytfXrsZwjz.UKVEsEPql6B33t4YzCktDDkJZ7nohFUVENtmCG1yKbT.e1UJAe.TBFRKc1Mr3asLNtboLiBSLiM.et4zVhYFKQ4zrL0OVUpaZYsqUFu5WfqVphsscMyZ0prWEqxl0SL73ea0T4gyDFS6M2.akD8VbXnDnrokYYPf502qR8ZlVIA2t9BgrOr8NJa9B5iuSYa7KoQTodZmSIZ5fG5Fu+YbhapzpMzs1stu6sOc2jxq5lbO2Mw5lzb2ZU33S4dsqG2b2ZU28Uc2uO5ta+wt6d0UsS9TdVwoCG1qBGeJ2rU2c2dU28Uc2u8c2u1o3tdxyVVpC8MCeiE5toBcrF4e7+b7wm9d+mEeYDbZ722MN2dBYx2frk6L9FQWBR0noK4QyT+q96PtkSP04LAyzAZeW6.e+SZ+mi9i+ZAcfW8j+2RGf8scBpLOO3suFogPLvmD+I0WtK95iwE.3SbBEuxQemFJaciXNf8viuc3MwsUzEsPuMc2EelK6UNNIayhlsN1KgNkWBcprD5TcIzY2kPmZKgN6MWcTWZdxoJTo2.iNGpOeiwgbxod56+.8+ZbukUC
-
@Steve-Mohican I'm not an expert in reading phase diagrams, but this looks alright to me, the phase is supposed to wrap around the frequency split points and it looks like a somewhat linear phase to me.
-
Yes I see what you mean. Regarding the frequency response, there is no problem. There are no any bumps or conflicts on the split points.
But the problem is the delay. You can see it in the phase diagram on the video. The wave look (similar to saw tooth) in the above video indicates the delay (a couple of milliseconds). Normally the phase diagram shouldn't look like that. It should look like almost a flat line.
Look at the below multi band compressor plugin. When the crossover frequencies are changed, there is no shifts like that. The wave shift is in the end of the line but it has so small interval. Also no change at all, even the frequencies are changed.
Normally, band splits don't cause any delay issues (causes phase issue). You can try any band split plugin.
NOTE: The phase diagram is
Inverted Raw Audio Signal
+
Processed Audio Signal
. The ideal should be a flat line. This is also called Null Test. Because if there is no delay, the signals neutralize (null) each other. The bump on the low frequencies in the below video is because of the high pass filter in the plugin, that's normal. -
@Steve-Mohican said in jlinkwitzriley phase issue:
But the problem is the delay.
If there is a delay, why don't you use the
Engine.setLatencySamples()
? -
@orange Because delay is not a fixed value. Unfortunately, it is constantly changing with band split frequency values :/