The timing / sync thread.

[Mr. Hyde]

I've noticed that when I record my parts often sound off time. Sometimes the timing of a chord or something will be so far off time that it sounds like delay. I have good timing so I know that I'm not playing that far off time. I've noticed that on a lot of the tracks I've listened to, that others here have recorded, that this might be a problem worthy of discussion.

I'd like to know if others here have experienced the following problems:

1. At the beginning of songs, or after a pause, the very first note I record is almost always off time. Often it takes me several takes to get the first note, at the beginning or after a pause to be on time. Does this happen to anyone else?

2. If I try to record a part straight through without breaks I find that my parts often drift in and out of sync, or change tempo, far enough that it starts to sound like delay sometimes. Does this happen to anyone else?

I think that this has something to do with the buffering going on between my hard drive, sound card etc...I've found that if I record in small sections at a time that the problem is less noticable in the finished track, but it makes it a longer process to record anything.

Any comments or ideas?

 

[rabies]

Hyde,

I have experienced this. To an extent. But I always thought it was my bad timing.

The begining issue is the worst. I always screw that up. I think we are suffering a lot because we don't record everything to a generic 'click' track.

We've discussed this issue on and off ever since I joined. We have sync issues, timing issues...the whole lot...the songs that I have worked on where we directly record against the original CD version (1 U Luv for example) are the tunes that I have had hte least problems with.

Other tracks supposedly were recorded against the original, but when you convert the original to .wav and play the track side by side..inevitably, by the end of the song, they have gone out of sync..often by about a second. Its a real problem. I don't have a solution.

...but a good thread none-the less..

 

[Mr. Hyde]

Rabies: I've also noticed that if If I try to convert the cd version of a song to .wav and the IMG mp3 version to .wav and insert them into my program that I usually can't get them to sync. It may be perfectly on time for a while but eventually it becomes pretty far off. I tried it with To tame a land most recently and I couldn't get them to be in time for very long. It is baffling to me.

I know that tape players, and record players can vary in speed from machine to machine but I thought that cd players were pretty much constant because of the format. I thought that If I play a certain copy of a cd here, and someone else plays their copy of the cd in another location, even on different equipment than I have, that we would get playback at the same tempo. Maybe that's not the case?

Maybe we are losing something in the process of the conversion we do to the files?

I know that it's been frustrating for me to record something and listen back and go wow how the heck did I get that far off time!

Maybe some kind of click track like you mentioned would help.

 

[rabies]

Maybe we could try some type of recording experiment.

...somebody create a click track..of ..oh...I dunno 90 bpm or something...for 3-4 minutes...Maidenman lays drums to it, bass player lays a simple line to it...two guitars record simple rhythm over it...all separate....all making sure to get PERFECT timing with the click track...but to the same original click file.

Then everybody submit their individual track and somebody puts it all together to see if each line matches the original click track on their end - and the rest of the instruments?

Think this may help?

We are paving new roads. We have learned that we can accomplish this collarborative recording over the net without a doubt...with some pretty good results. Now its all about adding polish..and one area we can work on is definately the timing/synching issues....mostly because they may be technically related.

 

[Mr. Hyde]

I think that it would be a way to help isolate what is going on.

We would have to use something really simple. No syncopation.

Something in a straight 4/4 timing. We could play a scale, in three part harmony, or octave/ harmony using the click track for reference, or just make up a simple rythem.

It would be boring as heck but it would be a way to see if the tempo and sync is being altered by something technical rather than us needing to be locked in a room with a metronome.

I think it's a great idea.

 

[jim]

Personally, I think everyone should be recording off the same reference track, be it a copy of the original with a 4 beat click intro added, or the bare drum track. Having different people just play to thier own copy of the original on CD is just begging for problems....

Also, don't forget that pesky bug we ran into before, where tracks were off-time based on conversion problems...

 

[Mr. Hyde]

I tried an experiment late last night. I recorded my metronome for about a minute at 100 bpm. Then I muted the track and recorded the metronome again for about a minute at 100 bpm on track two. Then I moved the tracks around to try and get them perfectly on time, with interesting results. The clicks wouldn't sync perfectly no matter what I did. There was always a slight delay between tracks, just enough to make the clicks sound like they were in stereo. I moved them where they sounded closest together and left them there.
Next, I tried to record a simple scale pattern along to the beat. I used notes from C major. C, D, E, F, G, F, E, D and repeated it over and over. I found that the first note I played ended up off time a little every time. I tried multiple takes and still the first note seemed to magically be behind the first tick. The second thing I found was that after a while some of the notes would seem to stretch and the next note after would be way off time. So far off that it would have taken serious effort for me to do it on purpose.
I'm trying to figure out what settings to change in my software etc...to see if I can correct that problem. I'd be interested to hear if anyone else tries the experiment and gets similar results.

I've always used the drum track or rough mix track to record my parts to. I agree that it gives us the best chance to be on time if we all use the same reference. I like to experiment and sometimes I'll take the IMG version of a song that I think used the cd for reference in the drum track or reference tracks, and I'll convert it to .wav, and then do the same with my cd version of the song and insert them into my program just to see how close the two versions are. I find that they won't usually sync up for very long, and by the end they are way off.

Did you have a disussion on this board about the conversion problem you mentioned? I don't remember reading about it. What was the problem, and did you find a solution?

 

[Eddy]

Originally posted by Mr. Hyde
Did you have a disussion on this board about the conversion problem you mentioned? I don't remember reading about it. What was the problem, and did you find a solution?

If I remember right, it was discussed on the old bb quite a while ago.

Also, if my mind serves me right, the conclusion was that the problem is the MP3-encoder/decoder applications used.

Gotta do some experiments myself, I've got an idea...

 

[rabies]

You could not get click tracks to sync? Wow....so there definately IS a problem that is not just our poor timing.. ...at least on your PC.

What did you use to record the click tracks? I'll try it on my side asap.

 

[Mr. Hyde]

Rabies: Yeah, the clicks would only get so close but they still sounded off far enough that I could hear them as two seperate tracks playing instead of it sounding like one click. I have a small battery powered metronome with a headphone output. I connected the headphone out to my soundcard and recorded it.
Then I tried my best to align the two clicks.

When I first started using this pc to record, I thought that my timing had just really gotten bad, but the more I record, the more I notice timing errors that aren't my fault.
(add the ones that are my fault and it's not pretty)

I can hear similar things in some of the other IMG tracks and I thought that maybe by starting this thread some things might get sorted out.

If it turns out that I'm the only one with this particular problem then I'll just have to deal with it.

 

[Eddy]

You definately aren't alone here. I've had the same kind of problems.

But I think I have a clue...

I'll do my experiment today, and give the results.

 

[Mr. Hyde]

I tried a similar experiment as the one I previously mentioned last night, but this time I used my keyboard to play a basic drum beat.

I couldn't get the two drum beats to be perfectly on time, at first, they sounded like drums with delay, but then I was able to get them close enough where it sounded like they were out of phase which happens at earlier levels of delay.
I found that my software isn't giving me an accurate visual representation of where the .wav's are aligned at the default setting. It has a magnification tool to view the files up close, so I zoomed in on them until they almost looked like sine waves, and found that they were still quite far apart. I aligned them while zoomed in and the phasing decreased quite a bit. The tracks were still out of phase slightly but it sounded like one drum track with a phasing effect on it. Much better than delay! I'll try the metronome experiment again and see if I can get them on time.

I still haven't figured out what is causing the audio to be off time at the beginning and to have tempo fluctuations.
I'm think I'm getting closer to finding the problem though.

 

[Mr. Hyde]

Progress at last! I was able to figure out that the timing was being thrown off by what seemed like a consistent amount. I started changed settings one at a time and noticed that I could make it so that beginning of my tracks were being thrown off by larger amounts of time, by changing these settings. I finally changed a setting called I/O buffer size, in my software. I tried smaller and smaller amounts 64, 32, 16, and recorded a small segment each time and the difference in timing became smaller and smaller. I thought that I had it figured out but then a new problem came in.
My buffer size apparently was too small because I started to have dropouts in my playback. I decided to try the other extreme with the buffer size. I set it to 512 and it worked!! For the first time the thing actually recorded pretty much exactly what I thought I played, and on time!! I've got my fingers crossed that it is the solution for the timing problems I've been having with my recordings.

Now, I think I can move on to the other timing/sync problems and start trying to work on those!

 

[rabies]

What might help here is knowing what software you use to record/mix.

 

[Mr. Hyde]

Well, the problem came back. It's not as bad as before, but it's started to delay the beginning of my tracks again. I still think I'm on my way to fixing the problem though. At least it's better now than it was. I'm getting closer to a solution I think.


Rabies: I'm using guitar tracks. Did you try the experiment on your system to see if you have problems? I know you've mentioned that your playing sometimes seem to be rushed or lag behind the beat . That's the same thing I'm getting, and a lot of it I can now definitely attribute to this pc.

 

[phlogiston]

Hi guys. I'm not a member but I check in occaisonaly just to see what you guys are up to. Feel free to disregard my opinion.

I think at least part what you're seeing (hearing) is an effect of signal processing, especially if you're recording off a track you are already playing on your computer.

What I mean is, when you hear a sound out of your speakers, it's been read from the disk into memory, been processed by your soundcard and converted to an analog signal and then spat out your speakers. If you play a note as soon as you hear it, it still has to be processed by your sound card, converted to digital and written to disk.

There's obviously a bit more it than this, otherwise you should be able to get two metronome tracks to sync.

I noticed this when I was playing around myself on and old PC. The recordings were about a second apart. When I got a new faster PC the problem was significantly reduced. (P 100 vs P3 733)

Commercial studios have Digital Signal Processors worth hundred's of thousands of dollars.

A good bet would be to buy a really good sound card and fast PC.

 

[Eddy]

Hi Pholgiston, nice to see some fresh blood around here!

Something that I have noticed is that when I do some recording under Winsucks (N-track), the sync error is almost alway present. But, when I've recorded under plain good-old DOS (GoatStudio) I've never had any syncing problems.

Gotta do some experimenting.

 

[phlogiston]

Originally posted by Eddy
Hi Pholgiston, nice to see some fresh blood around here!

Something that I have noticed is that when I do some recording under Winsucks (N-track), the sync error is almost alway present. But, when I've recorded under plain good-old DOS (GoatStudio) I've never had any syncing problems.

Gotta do some experimenting.

Thanks

And that could be because Winblows adds an extra layer to the process. The process is added so every program can access every sound card the same way (abstraction), whearas in DOS most programs have to use the sound card directly, so they process quicker. Sames as video cards. That's why DOS was a much better gaming platform until about DirectX 5.

 

[Mr. Hyde]

Hi, Phlogiston. I agree with you that it takes time for the notes to be processed and that might be causing the delay. My software creates an audio profile of my sound card and is supposed to help deal with timing problems. I have an Athlon 1GHz processor, and a really fast hard disk. So I don't think processing power, or my hard disk, is the problem on this system. I do have a generic built in audio card however. I can add a new card (with some troubleshooting)
but I can't be assured that it will cure the problem.

My software creates an audio profile of my sound card, and it is supposed to accurately determine the processing time that it takes for my card to process audio. I think that my software might be creating the delay by using the wrong profile for my sound card.

Also, this computer has Windows Me which might also be the cause of my grief.


Eddy: That's interesting, which version of windows/dos are you using?

 

[Mr. Hyde]

I decided to try out a new sound card again. I installed the new card and I have 14 days to evaluate it, find problems etc..., before I can't return it to the store. Installation went well, since I know what problems I encountered with the last card I installed and knew what to expect.

I immediately had problems recording with the card. I was getting a wierd static pop type sound in my audio. I changed some settings in my software and the static seems to be gone.

The sync problem also seems to have vanished. I recorded a basic riff and put a harmony over it, and everything seems to be in sync. I'm going to run this card through every test I can think of and make sure that everything works. Some things I can't test ,such as the digital inputs, because I don't have anything to connect to it. I'll test the midi and record some more to make sure this card is ok.

I have to make sure the remote control works!