I’ve just released a HUGE update for my very old (~6yrs) Cassette Deck 2 library. It’s very much like the recent update for Cassette Deck 1. I’m not going to show graphs with comparisons of the before and afters of the frequency responses this time, but the differences are very much like they were with CD1 (so you can just look at the graphs I posted for that update a couple posts back).
In this update, the library got reduced down from about a gigabyte, to ~65mb for the 96khz version. And it sounds many times better, even with the much reduced size! Here are the updates:
*Used the same method I used with Cassette Deck 1 to get a cleaner frequency response. The original programs had a ‘jaggy’ frequency response in the higher frequencies. This is due to wow/flutter speed fluctuations, which happen with any recording/playback medium that uses a motor. The fluctuations throw off NAT, which is the program used to create the impulses from the tone recordings, which results in the jaggy response. Sometimes you could hear it as a subtle ‘sizzle’ in the high end. Now the sound is perfectly smooth in the high end, like it should be.
*Edited/trimmed every impulse using techniques I’ve developed since originally releasing the library. This makes sure that the programs load and sound exactly as they’re supposed to. The old programs had some impulses that were out of sync, dynamically, especially in the higher dynamic steps. This would result in a very badly distorted sound if you drove the input too much. Now you can drive the programs a lot further and still have very useful results. The usable range of the sampled dynamics has been fully realized finally. This and the first update listed are HUGE. The effects are completely different now, they’re improved that much. You can get all sorts of interesting compression effects by driving the input differently, where in the old versions you’d just get bad distortions.
*Adjusted how the dynamics are handled internally, resulting in even more improvement to authenticity.
*Removed Attack control, the attack is fixed (and customized) internally to be as close to instant as possible with this type of Nebula program. Release control is still available, but with a much extended range. Now this control is actually useful and can be pretty great for fine tuning the dynamic response a bit. Increase it for a smoother result when driving the programs.
*Removed liquidity control. No need for it.
*Weeded out a few bad impulses that caused flaws in the sound of some of the programs.
*Weeded out a lot of harmonic impulses that contained only noise, no actual impulse (the lower sampled dynamic steps especially). These contribute nothing but still take resources. Removing them improves CPU use, RAM use, and the programs/vectors take up less disk space and load faster.
*Reduced kern amount for some of the programs because the higher amount just wasn’t needed for those programs. Further reduced resources needed.
*Shortened impulse length for harmonics since in my opinion they don’t need to be as long as the fundamental impulses. Reduced resources needed.
*Increased length of the fundamental impulses. In some cases this increases accuracy of the frequency response in the bass range, and maybe improves tone.
*All of the changes resulted in the CPU use being about 6-10 times LESS than the original versions, while the programs sound and work multiple times better than they did.
*Got rid of the different kern options, because the CPU use got reduced so much anyway. They aren’t needed now.
*Removing kern options allowed me to simplify the program names so they look nicer.
*The category in Nebula where the programs are found has been changed. Now they’re inside ‘TPE’ then ‘C39’ for 96khz and ‘C34’ for 44.1khz.
*Simplified the ‘Generic’ tape program. It was intented to be different from the others, and act as kind of a delay effect with a fixed repeating delay around 60ms. I cut out the delay because it didn’t work out so well.
*About 3 programs had their harmonics lowered in level a bit, because they would get too loud if those programs were driven hard. It didn’t sound very good.
*Updated the manual to finally have my logo (didn’t have it back then!), and kind of touched it up a bit here and there. Also updated the graphs of the frequency responses. Put in new tips.
I have as a goal, going through all of my older libraries and updating them, to be more in-line with my more current programs, in terms of how things run under the hood and how the programs are organized, etc. So this is the next one to get that treatment!
I expect to release the update… hopefully tomorrow. There are a lot of improvements. All the reverbs are using my current internal settings for reverbs (mainly talking about how the dynamics are handled). This alone probably should make a noticeable difference. I’ve moved the programs to a new category, looking to get some of my older programs better organized and have some different libraries grouped together.
I’ve edited the impulses so they don’t have excess length (as they used to), and weeded out a lot of impulses that just weren’t contributing anything (harmonic impulses where it’s just noise floor, no actual impulse). Also reduced sample count for harmonics that actually were above noise floor because there were just too many. These changes resulted in the reverbs being greatly reduced in filesize, which hugely reduced RAM use and in some cases also CPU use.
I simplified the library by removing 1k, 6k, and DIST versions, leaving only one version. I just don’t think the CPU use was that different between the 1k and 6k versions, or at least it isn’t now, with the better CPUs out there. Plus the reduced CPU use from weeding out the excessive impulses makes up for not having the 1k programs anymore. The DIST ones, I just don’t think were really necessary. Probably few people used them.
Last big addition- I made new alt versions for every reverb. Some are only subtly different, others are more noticeable. I think it gives more variety to the library.
Just sent out links for downloads for the update for Part 1. The updates for 2 and 3 will be released in the coming weeks.
Part 1 and 2 both get a new reverb- a bonus ‘fake stereo’ reverb I made by processing the original sources for the mono plates from both releases with a cool old Orban Stereo Synthesizer unit I have. All it does is use inverse comb filtering to produce a stereo output from a mono input. I discovered it can sound very cool for reverbs. So each of the formerly only mono plates now gets a new program with 6 selectable pseudo stereo reverbs.
I also fixed a bug in my internal program settings/programming that affected the release of the dynamic behavior. Probably nobody will notice unless they directly compare, but the fix probably makes them more accurate to hardware, in my opinion. It’s subtle.
Lastly I spruced up the names of the programs as they show up inside Nebula. Plates of Legend III will only be getting this and the previously mentioned bug fix- no new programs for it.
I stumbled across a method for greatly improving the frequency response of these older programs I made several years ago. Any recording medium that uses a motor for playback will suffer from some level of wow and flutter. This is fluctuations in the playback speed of the motor, which result in a ‘wobble’ of the audio’s pitch and speed. The problem is that this throws off the program used to sample things for Nebula- NAT.
When it tries to ‘deconvolve’ a tone sweep down into an impulse, the fluctuations caused by the wow/flutter are recognized as phase shifts in the frequency response by NAT. This leads to the tell-tale ‘jaggy’ response seen in any program sampled from a recording medium with a motor. With a good Reel to Reel deck the jaggyness can be very subtle, but if you zoom in with a good analyzer you should still see it. When it’s that subtle, it’s not a problem! Cassette decks usually have worse wow/flutter due to the motor speed being slower, so with these programs the jaggyness was actually audible in some cases, manifesting itself as a subtle ‘brightness’ added to the character, which wasn’t supposed to be there.
I’ve found a way to fix it, and the responses look and sound sooo much nicer now! There were *many* other improvements to the programs in the library, with an over-all result being, in my opinion, that this is basically a new thing. I think the potential for use has been greatly opened up.
Below are some before and after comparisons of some of the programs in the library, so you can see how the frequency response has been improved. These graphs don’t show how the dynamic behavior is also greatly improved in this library!
Old responses on left, new ones on right!
This is tape #3, Maxell UR. It’s a type 1 tape.
Here’s program #5A, a Maxell XL II which is a type 2 tape.
This is program #5B. It’s the same Maxell XL II tape, but this one was played/recorded at a slower speed while sampling.
Program #11A, a Maxell MX-S which is a type 4 metal tape. This one is great for a more subtle effect due to having the least high freq roll-off up to 20khz out of the set. The best program for that purpose. Still gives some of that tape crunch in the dynamics.
So I’ve just finished an update for both sets. Same thing is added to both. Each set has a stereo and a mono plate. The stereo plate in each set got additional variations such as a stereo swap version, a mono version that uses the sampled left channel impulses for both channels, one that does the same but using the right channel impulses, and another that uses a mono mix of both left and right sampled impulses mixed together for both channels which is a bit thicker than the other mono versions.
These aren’t something that can be obtained easily in your daw with channel routing. It might seem like it would be easy but it’s not, especially for the mono mix version. These are also more practical and convenient than using complicated mixing. They provide you with more, slightly different sounding (from the main stereo programs they are based off of) options for using in your mixes, and you may find that you prefer one of these alternate versions in some cases, if you try them out!
So, after a pretty long hiatus I’m back! Kicking off, let’s call it ‘phase 3’, with the Bee X-20 reverb, and I know I say this every time but this is definitely THE ONE. My best (released) reverb yet. Of course a lot of that is due to the hardware itself, but to recreate something as complex as that in Nebula, including the sampled damper control, is no easy task. But you can read more about that on the product page (just check it out under the reverbs category)!
Hopefully this is just the start of another string of cool Cupwise Nebula products and I can get at least a few of the things that have been sitting unfinished (or partly finished like this release, serving as my own personal super secret cool reverb for a little over a year now!) on my hard drives for up to years now, “out the door”, before fading back into the mist….
Incidentally, this also happens to be the first new release put out on my website since the site got updated!
Back about mid-way through 2014 I had a pretty extensive guide to sampling for Nebula (using the NAT sampling program) published in Sound on Sound. It’s aimed at beginners and I tried to de-mystify a lot of the notions about sampling in it. At a glance the guides might look pretty complicated but I think they break it down into pretty solid step-by-step instructions which should help anyone get something useful. It covers 3 types of programs- Preamps, Reverbs, and non-dynamic Multis. Multis are what anything that’s sampled with an adjustable control which is sampled in multiple positions is called, which includes equalizers for example.
I think I also gave a pretty good idea of some of the major limitations to the platform. There are also some example programs and custom NAT sampling templates which are from my own personal set, which you can download at Sound on Sound, as supplemental material for the article. The programs are from the examples I made while doing the guide, as a walk through.