| len-dt | ailo, Just did an a/b with high precision timers accessible or not. | 01:02 |
|---|---|---|
| len-dt | The apps do need to have access. I recorded the synth demo on qtracktor and then played it back to the synth | 01:03 |
| len-dt | without the hpt, note timing was all over the place. | 01:04 |
| len-dt | with hpt it was much better. | 01:05 |
| len-dt | One of the things the script did not check is hpet/max-user-freq and rtc0/max_user_freq. stock is 64 | 01:08 |
| len-dt | rosegarden suggests 3072 | 01:10 |
| len-dt | this may tighten the timing even more. Unless RG is the only app that tries to go that high. | 01:10 |
| len-dt | micahg, did our packages get released? | 03:30 |
| micahg | len-dt: sorry, didn't get to it this morning | 03:31 |
| micahg | still on my list though | 03:31 |
| len-dt | Ok I won't test it then :-) | 03:31 |
| === shnatsel is now known as SergeLion | ||
| === shnatsel is now known as SergeLion | ||
| len-dt | ailo, I have been thinking about the testing we have (ok I have) done with the high rez timers. | 20:21 |
| len-dt | While it is obvious just listening that there is a difference, The mention of maxfrequency makes me think we need something more precise. | 20:23 |
| len-dt | The usermaxfreq is 64 right now but 3k or so is suggested by at least some people at rosegarden. | 20:24 |
| len-dt | They do say to try 1024 and 2048 though to see what works best. | 20:24 |
| len-dt | I am thinking I will not be able to hear the difference very well just with my ears. | 20:25 |
| len-dt | So, I am thinking to make a sequence on hydrogen that is very specific with it's timings... no swing. | 20:26 |
| len-dt | Run it out through the midi port and back in to record on qtracktor. Then playback qtracktor through the midiports and rerecord on tractor again. Then visually look at timing with different settings for max freq. | 20:28 |
| len-dt | ailo, does this sound like a valid way of testing? | 20:28 |
| ailo | len-dt: In my experience the timing issue is already significantly visible when you record only once | 21:23 |
| ailo | len-dt: The timer thing must be one of those things we most probably should fix for 12.10 | 21:23 |
| ailo | And perhaps call that a bug for 12.04 as well | 21:24 |
| len-dt | ailo, yes it is, but when the HPT gives access to apps it gets better. | 21:24 |
| len-dt | I am thinking about maxuserfreq testing. | 21:24 |
| ailo | I'm still occupied with this other thing, but in one week I would like to look through everything that you have done, and see what I can do on top of that | 21:25 |
| ailo | And start putting stuff up on the wiki, etc | 21:25 |
| ailo | Also, I want to create the -controls | 21:25 |
| len-dt | Good | 21:26 |
| len-dt | The only reason for the second record step was so I could see where the notes line up with the beats. | 21:26 |
| ailo | len-dt: You should be able to compare each instance of the beat | 21:27 |
| ailo | Cut one out, and paste it under another | 21:27 |
| ailo | Then compare the two | 21:28 |
| len-dt | ailo, fixed delay is ok, variable is not | 21:28 |
| ailo | For me, there were major differences between each bar | 21:28 |
| len-dt | a stock install has problems yes. :-) I could here it easily too. | 21:29 |
| len-dt | s/here/hear | 21:29 |
| len-dt | With access to HTP I can't put my finger on it, but sometimes when a lot was happening it seemed a bit "muffled" even with HPT access. | 21:31 |
| len-dt | I am thinking increasing the max freq might help that. | 21:32 |
| ailo | I had no problems with jack midi, using my firewire HW | 21:32 |
| len-dt | Stock seems to be 1/60 sec. | 21:33 |
| ailo | Oh | 21:33 |
| len-dt | ailo, with 12.04? | 21:33 |
| ailo | No, but I don't think it matters which OS | 21:33 |
| ailo | Same settings | 21:33 |
| ailo | jack midi is what you expect from midi | 21:34 |
| len-dt | hmm | 21:34 |
| ailo | This is what I told you about 3 months ago, or when it was, when we were discussing these things | 21:35 |
| ailo | Firewire devices use jack midi for the hardware | 21:35 |
| ailo | They don't use alsa midi at all | 21:36 |
| len-dt | Yes, I remember. It is just that there seems to be no trouble midi in to midi out going through alsa, just when an app is recording/playing. | 21:36 |
| ailo | When else would you use midi out to a HW device? | 21:37 |
| ailo | Apart from controlling levels | 21:37 |
| len-dt | kb to synth either internal or extrnal | 21:37 |
| len-dt | or through a midi filter. | 21:37 |
| ailo | The issue was always on playback | 21:38 |
| len-dt | If the problem was with the ports it should show up then but doesn't | 21:38 |
| ailo | I never investigated the other way around. To do that, you either need a good ear, or an external sequencers | 21:38 |
| ailo | sequencer* | 21:39 |
| ailo | Perhaps using jack midi out, and record with alsa midi | 21:39 |
| ailo | I could do that in 1-2 weeks time | 21:39 |
| len-dt | That was how I tested at first. It was when I realized the problem was with internal SW that I started using that, | 21:39 |
| ailo | len-dt: You didn't use jack midi out, and record with alsa midi | 21:40 |
| ailo | len-dt: I'm saying, this would be a way to test recording midi | 21:40 |
| len-dt | I don't have jackmidi ports. | 21:40 |
| ailo | I do, but not at home at the moment | 21:40 |
| len-dt | mine are all alsa. | 21:40 |
| ailo | But I will try this in 1-2 weeks | 21:41 |
| ailo | Many people won't hear that the recording is off, cause they expect themselves to not make timing errors | 21:41 |
| ailo | I mean, the other way around :) | 21:42 |
| ailo | They expect to make timing errors, human timing errors | 21:42 |
| ailo | So, when they get it, it sounds normal | 21:42 |
| ailo | With an external sequencer, you could do the same thing as I propose | 21:42 |
| ailo | hmm, or maybe not | 21:43 |
| len-dt | That is why I want to start with a quantized track. | 21:43 |
| len-dt | ailo, alsa midi just forwards what the port does, is that correct? | 21:44 |
| ailo | len-dt: In my experience, the problem only occurs once the midi is sent through a port | 21:45 |
| len-dt | does Jack add some timing info as well? | 21:45 |
| ailo | timing info? I only used my ears | 21:45 |
| ailo | And looked at the wave files | 21:45 |
| ailo | Jack midi was on the spot | 21:46 |
| len-dt | The old roland 401 used to add the amount of time since the last event | 21:46 |
| ailo | I was doing a recording, which is why I needed something that worked. alsa midi did not, so luckily I had a FW devices | 21:46 |
| len-dt | So even if the SW took a while to get to it, it was aware of when it happened. | 21:46 |
| len-dt | I am wondering if jack does something similar | 21:48 |
| ailo | No idea | 21:48 |
| ailo | Ok, gotta go. Need to get this ship sailing | 21:48 |
| len-dt | I guess it doesn't matter, there are lots of USB midi ports out there gotta make em work. | 21:49 |
| len-dt | Bye now. | 21:49 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!