Intermittent garbled sound

[dbp2k]

Set up: Ubuntu Studio 13.10 "out of the box" install. Dell XPS Studio m1340 laptop. Just did a clean install this weekend - not an upgrade.

Background: I wanted to add beats to an audio track I had of simple vocals and guitar (wav file).

Approach: Add wav as a sample track.

Symptoms: Before adding any beats or making any changes, I try to play the song in Song Editor just to confirm audio is OK. Sometime it sounds fine. Sometimes it sounds horribly garbled. When it sounds garbled, I can't find any setting to fix it so I just exit LMMS without saving and restart it. Sometimes it sounds good after restarting, sometimes garbled.

On one of the sessions where it sounded good, I tried adding a beat track. Sounded OK for a few minutes but then it sounded all garbled again.

Assuming this is a PEBCAK problem, I tried researching but found no answers.

BTW - I am interested in pointers on technique to match BPM for an audio sample recorded without a metronome.

I hope I can get past this newbie issue. The UI is good and the feature set looks good. I'm just having a lot of unexplained start up problems for a new user.

[musikbear]

Most often 'garbled' sounds cames from a 'wrong' or rather insuficient supported choice of sound-device. Most will do best with the choice SDL
so in
Edit| Settings| Sound
check if you have SDL as default

-btw You explain your issue well.

BTW - I am interested in pointers on technique to match BPM for an audio sample recorded without a metronome.

You need to 'tap' that yourself. You can find small proggies to aid you. If you have a demo of fls, there is a beat-tabber, but its a somewhat hefty pack to install, just for a beat tap proggie

[diiz]

Personally, I'd skip the userland sound systems and just let LMMS talk directly with the kernel sound drivers

1. Select ALSA as sound device

2. In the "device" box, enter the designation of your soundcard, which will most likely be hw:0,0 if you only have the one *

3. Restart LMMS

The downside is that you can't get sound output from any other software while LMMS is running, but the sound will work best this way.

*) - to find out your soundcard hw:x,y designation, look at these instructions -> http://lmms.sourceforge.net/wiki/index. ... ts_in_LMMS

The other option is to use JACK, which you should have preinstalled since you're using Ubuntu Studio, but there's not much advantage in doing so unless you want to route sound to other programs. With JACK, you have to set the settings right and start the JACK daemon before starting LMMS (usually you do this with qjackctl). For the settings, you have to set the sample rate to 44.1khz, because currently LMMS doesn't seem to be able to output any other sample rate.

[nathan-triplex]

The sound problem will most likely be your audio output (in the settings).

Getting the right BPM I often add the sample as an instrument and match the BPM within LMMS.

[diiz]

Forgot to mention, one other advantage of using JACK is that you'll be able to run Audacity at the same time, if you use pure ALSA you'll have to switch between them (close LMMS first before opening Audacity). It's convenient for testing new VST synths, to see if they render properly in LMMS (some do not output over 44.1k sample rates, not sure if it's a problem with LMMS or the particular VST's): you can render some test piece and open it in Audacity, then compare it to the sound that is produced live in LMMS.

Of course, you do still get the best performance with pure ALSA, but you have to be sure that LMMS is routed directly to ALSA (by using the hw:x,y trick), otherwise it will use PulseAudio even though it says ALSA (PA is designed to "intercept" ALSA output) and PulseAudio just plain sucks for any kind of audio work.

I also just noticed that you don't have to run qjackctl (or similar) before opening LMMS, it seems LMMS is able to start JACK by itself. That's nice.

[dbp2k]

Thanks for the tips. I checked sound settings and it had picked ALSA by default. I changed the setting to SDL and the garbling symptoms seems to have gone away (at least for now). I also found a couple of "beat" tools very easily - thanks for the tip.

Re matching audio with beats .... it is proving rather challenging. The audio sample was a live performance done without a metronome. It is not precisely keeping to a constant BPM. My first attempt was to place beats using the piano roll to match the audio. Very tedious and in some cases, I'm still not able to really "nail" it within a 16th note granularity. I'm starting to think I would be better off laying down an accurate beat and cutting and pasting the audio to fit (at least conceptually, although I don't see an intuitive way to do this using LMMS alone). I'm sure this is a common problem and there must be tried and true techniques for doing this - I just haven't found them yet.

[diiz]

Re matching audio with beats .... it is proving rather challenging. The audio sample was a live performance done without a metronome. It is not precisely keeping to a constant BPM. My first attempt was to place beats using the piano roll to match the audio. Very tedious and in some cases, I'm still not able to really "nail" it within a 16th note granularity. I'm starting to think I would be better off laying down an accurate beat and cutting and pasting the audio to fit (at least conceptually, although I don't see an intuitive way to do this using LMMS alone). I'm sure this is a common problem and there must be tried and true techniques for doing this - I just haven't found them yet.

Sadly, I don't think there is a good way to do this with LMMS, at least yet. LMMS is fine for synth/sample based music, but if you want to use long audio tracks, and especially if you want to do beat matching to a varying tempo, LMMS is probably insufficient to the task, due to several issues:

- There's no way to start the playback of a sample track from the middle, you'll have to rewind back to the start of the sample track every time you play (gets old really fast with long tracks)
- There's no way to do a proper tempo map - you can automate bpm, but you have to do it by manually editing the curves, and the interface does not respond to it, by which I mean it will not "synchronize" sample tracks to the bpm track, so you won't be able to match the sample track to the tempo by looking at the waveforms, either

There's reason to believe these deficiencies may be fixed some time in the future, but there's no telling when that's going to be. For now, your best bet is to try something like Ardour.