I’m not sure where to post this but let’s give it a try here :
I am using hydrogen 0.9.4 svn for some drum beats. I create tracks in ardour to record some of the HG instruments. So far so good. But I notice that sometimes, a couple of tracks have noise in them. It does not come from my hardware (RME multiface II), that’s definitely not the source of the noise. The ardour track inputs are strictly connected to HG’s instruments outputs.
See for yourself on this snapshot : I armed some tracks for recording in order to make the noise visible. You can see at least two tracks where the noise is at an unacceptable level. I am sure of my routing, it is very simple. I suspect the problem comes from hydrogen but again, I cannot claim this with 100% confidence (I hope the image below is fast at showing up). The result during playback is quite awful to hear …
I suspect your levels are too high coming from Hydrogen. Have you tried setting lower values with Hydrogen’s faders ?
Also, it looks like your tracks are mono. Are you adding together the stereo outputs from each track in Hydrogen ? That might also contribute to distortion if the summed amplitudes are too high.
I don’t have the session opened in front of me (the snapshot was taken at a very early stage of a project) but I can say this :
1- I am sure the HG faders were not too high
2- if my HG tracks are mono, it’s because I checked the HG-track-L and HG-track-R outputs beforehand and to my surprise, the track-R outputs have nothing. When I start a project, I check sources through jaaa (JAck Audio Analyzer) to see what the signal’s relevant part of the freq spectrum is, and this great little tool showed me that all HG tracks have no signal coming out through the right output of each and every track. I didn’t have time to dig it but the main HG outputs (out1 and out2, the mixed down outputs) are on the contrary doing fine. Either something is fishy with the svn version of HG or my HG setup is flawed. But in any case, I only use the track-L output (I don’t sum up -L and -R since -R has nothing in it).
But I will come back to that as soon as I have time.
Hmm… I tested the Hydrogen-Ardour combo in 64 Studio, with Ardour 2.1 and Hydrogen 0.9.3. The stereo track outputs did in fact record in stereo to Ardour, with at least one interesting anomaly (though I may have a track routed incorrectly). The recorded levels were hot, I needed to adjust faders in Ardour to avoid clipping.
I’ll try the same test in JAD today, it includes the same version of Hydrogen that you’re using.
weird … but an easy way to convince yourself that something is odd, is to directly connect the HG track outputs to your speaker system (playback_1 and playback_2) with qjackctl if you like. Nothing comes out of the right component of the instrument track. It looks like HG sumps up left and right in the left channel because the level is too high there. Probably a HG bug …
… I use hydrogen 0.9.4-svn. Could be an issue in the SVN code. I will try 0.9.3 with a dummy session.
I had the idea it might be the filter, envelope or pitch settings on particular hydrogen sounds causing this, as that would affect only some of the outputs.
I gave it a go with various settings, and stereo+mono tracks, but I don’t get the noise while everything is stopped.
The stereo ouputs also seem to be panned centreish, I don’t get it louder in one side.
Does the noise always follow the voice output assignment?
I used Ardour/GTK 2.0.5, jackd 0.107.2, Hydrogen 0.9.3
Hydrogen continues to pose problem. But their website being down, I cannot post on their forum (I did that but the HG forum is not very active).
I repeat myself : the noise you see when you arm tracks for recording is showing up even if the transport is not rolling. There is a noise source somewhere that has nothing to do with hardware and I believe it comes from hydrogen but if someone had similar experience, I would think it’s at least not due to my setup.
new snapshot (with transport off) :
I tried HG-0.9.3, no problem here (no noise, and stereo tracks OK). So the dev version is still very buggy. The drawback with using 0.9.3 is that I have to resample ALL drumkits to 96000Hz. Fortunately, I have a little script for that …
Some update here : I finally had the time to look at the hydrogen code yesterday and found an obvious bug in the 0.9.4 SVN version. The right channel audio is definitely routed to the left channel (stupid copy/paste of the left channel code I suppose). I notified the hydrogen forum.