7704A
Well-known member
Mostly just venting/ranting. I've been going down a rhythm/timing rabbit hole as of late, and right now I'm focusing on nailing the click to the wall. A few days ago I realized that a big issue was that I didn't know what dead on sounded like, so I dug around the internet and found some advice on hearing it. Great. Now I'm largely within 10ms, with a significant fraction of hits being within 6ms or less of the click transient. Improving beyond that has proved challenging though. I think I hear things one way, and when I check the recording of me and the click, I'm off. So the last day or so I've been trying to learn how to hear the difference between the click and me for differences below 5ms without a whole lot of luck. I'll practice to the metronome, feel like I've locked things down tight, then when I measure with the computer, I'm off by a couple milliseconds. This has made it harder to train too since the difference in what I hear and what I measure makes it hard to know if I've actually hit the click.
So far I've been assuming that the latency in my recordings is the same as the buffer on my 1st-gen Scarlett 18i20, about 1ms, so I've been expecting a 1ms difference but not more than that. I actually switched to a different click noise with a sharper transient because the old one peaked several ms into the sound, and my hits were sorta centered several ms late. So I figured that I was tracking with the peak of the transient, not the start, and that it would be easier to sync to the start of the click if the wave form was more like a brick wall.
I finally decided to check the latency for my headphones that I'm using to both listen to the click on and to record myself during tests. Note that I'm not using the 18i20 at all for this, I'm just going into my computer's headphone jack. Because I couldn't get the headset mic right next to the headphone speakers (about 6" away) my measurement is probably about 0.5 ms longer than the actual delay. What I got is about 3 ms of latency between the click being recorded, and the click traveling through the headphone speakers, into the mic, and back to the DAW. Bigger than I thought.... I started playing around with some recordings of myself and the click, shifting things back and forth by 3 ms. Yup, definitely enough to change how I hear things. I then do a couple recordings of a few clicks each, noting how they sound to me than comparing my prognosis to what the DAW says after latency correction, which confirms that there is a definite difference in how dead-on actually sounds, and how I've thought it sounded based on my prior measurements. Great. Because of this, I've been trying to play a few ms early while at the same time trying to play what I hear as dead-on, which obviously doesn't work. That explains my confusion over hearing the click...
The last thing I did was double-check the old metronome click sound. I pull up those takes and check the spots where I was centered over the peak and not the start of the transient and... the peak is 3 ms after the transient. So I was syncing to the start of the transient.... And my head's just been getting screwed over by not accounting for the latency when measuring. Gah!
Well, back to the woodshed. I got some unlearning, relearning, and a lot of practice to do. While it appears that I'm more accurate than I thought, consistency is still not where I want it to be.
Moral: Check latency.
So far I've been assuming that the latency in my recordings is the same as the buffer on my 1st-gen Scarlett 18i20, about 1ms, so I've been expecting a 1ms difference but not more than that. I actually switched to a different click noise with a sharper transient because the old one peaked several ms into the sound, and my hits were sorta centered several ms late. So I figured that I was tracking with the peak of the transient, not the start, and that it would be easier to sync to the start of the click if the wave form was more like a brick wall.
I finally decided to check the latency for my headphones that I'm using to both listen to the click on and to record myself during tests. Note that I'm not using the 18i20 at all for this, I'm just going into my computer's headphone jack. Because I couldn't get the headset mic right next to the headphone speakers (about 6" away) my measurement is probably about 0.5 ms longer than the actual delay. What I got is about 3 ms of latency between the click being recorded, and the click traveling through the headphone speakers, into the mic, and back to the DAW. Bigger than I thought.... I started playing around with some recordings of myself and the click, shifting things back and forth by 3 ms. Yup, definitely enough to change how I hear things. I then do a couple recordings of a few clicks each, noting how they sound to me than comparing my prognosis to what the DAW says after latency correction, which confirms that there is a definite difference in how dead-on actually sounds, and how I've thought it sounded based on my prior measurements. Great. Because of this, I've been trying to play a few ms early while at the same time trying to play what I hear as dead-on, which obviously doesn't work. That explains my confusion over hearing the click...
The last thing I did was double-check the old metronome click sound. I pull up those takes and check the spots where I was centered over the peak and not the start of the transient and... the peak is 3 ms after the transient. So I was syncing to the start of the transient.... And my head's just been getting screwed over by not accounting for the latency when measuring. Gah!
Well, back to the woodshed. I got some unlearning, relearning, and a lot of practice to do. While it appears that I'm more accurate than I thought, consistency is still not where I want it to be.
Moral: Check latency.
