Scoring Central

Full Version: VCSL
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
I ordered another microphone, did some reading about NOS stereo recording technique, and made an attempt at sampling the clarinet. My room still doesn't have any proper sound treatment, so I stood behind the mics and looked down the "line of sight" on each of them, then hung up blankets where the cardoid pickup would be the strongest. I played the instrument about 0.8 meters away from the mics, and had the gain on both of them dialed in at about 60%.

https://www.dropbox.com/s/m8iz33kbo0uomg...7.wav?dl=0

Sam, will you give me some feedback on this? It would be very helpful to know if there are any adjustments I can make that would improve the sampling quality. It occurred to me that I know a lot of people who play different instruments, so I can possibly contribute quite a bit to VCSL if I can get satisfactory recordings.
(04-17-2018, 03:24 PM)Michael Willis Wrote: [ -> ]I ordered another microphone, did some reading about NOS stereo recording technique, and made an attempt at sampling the clarinet. My room still doesn't have any proper sound treatment, so I stood behind the mics and looked down the "line of sight" on each of them, then hung up blankets where the cardoid pickup would be the strongest. I played the instrument about 0.8 meters away from the mics, and had the gain on both of them dialed in at about 60%.

https://www.dropbox.com/s/m8iz33kbo0uomg...7.wav?dl=0

Sam, will you give me some feedback on this? It would be very helpful to know if there are any adjustments I can make that would improve the sampling quality. It occurred to me that I know a lot of people who play different instruments, so I can possibly contribute quite a bit to VCSL if I can get satisfactory recordings.

First off, don't be afraid to record with more gain- ideally the loudest recorded sounds should be around -6 dB or so (so there's plenty of extra space for clipping). Too much lower and you risk introducing noise from the equipment when much gain is added.

I noticed that the two channels were not quite at equal volume. This is a very tricky job when recording with analog/manual gain knobs on the interface. The easiest way to balance two preamps is to have some sort of signal generator (e.g. an electric guitar/piano/etc) that you run into one preamp and play a tone, then take note of the specific volume on the monitoring software/DAW. Then, switch to the next preamp and add or remove gain until it matches the same volume. This should be done when recording with a microphone pair. Note that, especially with cheaper microphones and cheaper preamps, the tolerances and the exact placement of the knobs means less about the actual gain being applied- my Zoom H6 for example (with preamps marked 1-10) differs by almost as much as a number when properly calibrated like this.

I also noticed a drone-like sound in the background, likely a fridge or air conditioning unit. Luckily something like this is easy to remove with a Notch Filter tool (like a super-duper small-bandwidth EQ). Combined with a pass through the denoiser, this gives a very reasonable noise floor of about -80 dB (down from the mid-50's).

I'm not sure what you mean 'behind the mics'? You want to be close to the middle line between where the capsules are pointing.

Anyway, here's the processed version I did if you want to have a listen or a go at cutting them-
https://s3.amazonaws.com/SamulisRandom/s...cessed.rar

Best way to figure out the tone/room sound you want is to try playing the instrument at a few different positions/distances until you come across one you really like. It may be closer or farther than recommended. Also don't be afraid to try different setups, such as spaced pair or ORTF. It doesn't matter how you do it, so long as it sounds the best and has reasonably low noise. Smile
Thanks Sam. When I said "behind the mics", I meant in order to be assured that the mics were not aimed at hard reflective surfaces, I went around behind the mic stand, closed one eye, and looked straight along the length of each mic, like sighting the aim of a gun (for lack of a better analogy). That gave me an idea of where to hang up some poor man's sound treatments.

I did notice after the fact that the gain knobs onmy interface were not even, but as you suggest I should really calibrate the gain on both inputs.

I'm fairly certain that the drone is the computer. The heat was turned off, and the refrigerator is behind a door, down the hall, and around the corner. My video card is kind of noisy, eventually I want to replace it with a fanless one.

I think I'm still a bit too close to the mic, I want to try around 1.5 or 2 meters. Also I noticed thatI need better playing technique in the upper register. Some of those notes had a wheezing noise of air escaping from bad embouchure.
(04-18-2018, 02:59 AM)Michael Willis Wrote: [ -> ]Thanks Sam. When I said "behind the mics", I meant in order to be assured that the mics were not aimed at hard reflective surfaces, I went around behind the mic stand, closed one eye, and looked straight along the length of each mic, like sighting the aim of a gun (for lack of a better analogy). That gave me an idea of where to hang up some poor man's sound treatments.

I did notice after the fact that the gain knobs onmy interface were not even, but as you suggest I should really calibrate the gain on both inputs.

I'm fairly certain that the drone is the computer. The heat was turned off, and the refrigerator is behind a door, down the hall, and around the corner.  My video card is kind of noisy, eventually I want to replace it with a fanless one.

I think I'm still a bit too close to the mic, I want to try around 1.5 or 2 meters. Also I noticed thatI need better playing technique in the upper register. Some of those notes had a wheezing noise of air escaping from bad embouchure.

With a cardioid, the elimination of noises to the side is typically quite small, perhaps 3-6 dB, although the characteristics of the mic may change significantly to the sides (such as picking up less of the highs). Even with the "notch" in the back, the reduction may be as little as 12-24 dB and it is very directional. It thus almost more important to make sure the mic faces away from noise sources (windows, computers, fridges, etc.) than towards the "right" place.

That being said, putting up some blankets is a great idea. Anything you can do to dampen the room is great if you find it "overly active", especially with high frequencies.

Farther back may be a good idea. I did notice it sounded a little close.
Added some new instruments-

Dan Tranh - Sourced from 'Dan Tranh'; mono. May replace in the future if it can be re-recorded in proper stereo
Bell Tree - Legacy
Gong 2
Nepalese Bells
Hand Chimes
Updated Sleigh Bells samples - Found a few samples that were left out
Tambourine 4 & 5 - Tambourines from Miscellania II
Tubular Bells 3 - The original test sampling for the original Tubular Bells; in .ogg format
Timpani 2 - Sourced from 'VS Timpani'
I've been working on some scripts for automatically generating sfz files.  If you'd like me to, I could polish them up enough to generate initial files for all these instruments.  Since you've used consistent file names, it would be easy to generate velocity layers, round robins, and keyswitches automatically.  It also can correct for errors in tuning and remove silence at the start of samples.  (You can see both of those at work in the SSO solo violin 2, where a lot of the samples are badly out of tune, for example https://github.com/peastman/sso/blob/mas...ibrato.sfz).  And I've been trying to automatically correct inconsistent volumes, but so far I haven't come up with a robust way to do it.  But I could spend a bit more time on it and try to make it work.
(04-20-2018, 04:01 PM)peastman Wrote: [ -> ]I've been working on some scripts for automatically generating sfz files.  If you'd like me to, I could polish them up enough to generate initial files for all these instruments.  Since you've used consistent file names, it would be easy to generate velocity layers, round robins, and keyswitches automatically.  It also can correct for errors in tuning and remove silence at the start of samples.  (You can see both of those at work in the SSO solo violin 2, where a lot of the samples are badly out of tune, for example https://github.com/peastman/sso/blob/mas...ibrato.sfz).  And I've been trying to automatically correct inconsistent volumes, but so far I haven't come up with a robust way to do it.  But I could spend a bit more time on it and try to make it work.

Wow, that script sounds very impressive!

You're more than welcome to work from the samples, just make a fork on Github, then sync that to desktop and build to your delight. Save and push the changes, then when you're ready, I can do a merge to bring it into a new branch on the main project, or you can keep it on your end and I can link to it. Only problem is you would need to update from the main fork every time I upload new samples.

Alternatively, I can add you as a contributor and just give you your own branch to work from or something.
No problem, I'm happy to work through pull requests. Give me a week or two and I should hopefully have something preliminary to show.
(04-20-2018, 04:49 PM)peastman Wrote: [ -> ]No problem, I'm happy to work through pull requests.  Give me a week or two and I should hopefully have something preliminary to show.

I should note that the consistency isn't as global as it may seem. There are some which don't comply well with the standard. If there are any which you need batch renamed, let me know and I can process them on the main branch.
Here's a few inconsistencies I've come across so far regarding naming.  My current thought is to split the filename between underscores and try to interpret each piece, so that depends on having a consistent syntax for each piece.

Most instruments indicate the note by name.  But a few of them indicate it by number instead, such as the pipe organ.

Many instruments indicate velocity layers by dynamic markings like "mf" and "ff".  Others use the specifications "vl1", "vl2", etc.  The electrophones indicate it with the numbers 53, 92, and 124 (the velocities they were generated for?).  The glockenspiel uses "soft", "medium", and "loud".

Round robin is indicated in a variety of ways.  Most often it's by "rr1" and "rr2".  But sometimes it's by adding the digit after the dynamic marking, such as in the crash cymbals.  Or it can be a digit all by itself, like with the cajon.
Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16