Mplay wrote:It's explained by Michael himself on the tape booster + manual on cdsoundmasters site. Tweaking the in and output on the glob page in the edit menu is really useful for getting awesom results
i am pretty sure from talking to Michael this is adding harmonics and not compressing. However i could be wrong so hopefully he can add to this.
Since the compression aspect in conjunction with the character aspect of equipment is still not quite 100% with Nebula there are other ways for percieved loudness increase. TB+ cleverly uses one of those ways by means of increasing correlated harmonics dynamically to give the impression of louder and fuller signals.
Hi there! Yes, everything stated here thus far is correct. What TB+ is doing is filling a void in the manner that Nebula allows, making it extremely realistic to the way that tape increases perceived volume. Nebula handles saturation effects in a separate category from compression effects. Compression is based upon measuring complex changes in the attack, release, threshold, ratio, and even some other programmable parameters in the engine itself, but the more subtle the type of source compression, the less it gathers it in the same manner. To keep it simple, if it is very subtle, constantly adapting deep leveling/limiting, it isn't going to come across correctly. TB+ is completely based upon the harmonic distortion aspect of tape, but because it works very well and Nebula does such a great job with this, it helped in research to define where the line between saturation and compression actually takes place in different machines.
I'll try to describe this in short summary, because the results come from an absolutely gigantic amount of research and testing.
This goes back to around 2005 when I started the tests specifically regarding tape. By the time that Giancarlo and I started writing each other about the harmonics found in op-amps and transformers, probably around the end of 2006, he was already formulating the complex processes that the Neb engine would handle correctly and I had devised a separate testing process that is similar to the concept of nulling, but it involves many more stages and requires testing based on sound rather than numerical data. By the time that R2R was ready as a concept, I was convinced that the only truly accurate and fair way to provide THE tape sound digitally, was to break the sound into 3 separate stages.
The first is to address the more important literal machines individually. This is what R2R represents. Because there is a range of sound that comes along with using these machines at their best settings, Nebula handles this flawlessly. This means, that over a relatively wide dynamic range, that the multi-dimensional character of many of the best sounding machines could be captured and reproduced and the user could choose, from a single interface, which machine they want, what speed to run it, and even in some cases which tape stock they think sounds best for a particular track and even if the character of the machine calibrated for hotter signals gave a different spectral response than if calibrated for a lower level signal.
Therefore, R2R represents "stage 1"- accurately recreating the character of specific machines, within the most common specification, in multiple dimensions that affect the sound the most, with an accuracy that comes directly from the actual machines themselves. I believe that R2R achieves this. I believe that it uses the resources of a single instance of the Nebula engine to get every bit of subtle character out of the real machines and I'm very please with this. It is particularly important because anything added to this relies on its accuracy to work correctly.
Stage 2 is TB+. Once again, using a single instance of Nebula, placing all attention on what the majority of tape machines share in common regarding anything that I can measure OTHER than spectral change that only relies on frequency (not phase, not flange changes, and not undefined generation of energy which is unintentionally being enumerated). Most machines share a common range of how loud they get before saturation becomes more and more apparent, and most machines share many similarities in the exact harmonic saturation structure at very similar volumes. When eliminating, for instance, the difference between the frequency plot in a cranked Studer versus a cranked Teac cassette running TDK or Maxell, amazingly the drive and compressing characteristics align to some commonalities in all machines. What I didn't know that I would discover is that there was a lot more going on in the trickier, more subtle areas of this sound regarding which aspects were coming from compression and which ones came from saturation.
I figured that the 2nd stage, TB+, would only account for more of the higher gain drive characteristics that came from getting the exact harmonic nature to line up correctly to the dynamic balance of what was feeding it. I expected this to work well, but it actually helped to verify other things I didn't know, which is that the combination of the specific tape machine combined with the specific amount of saturation it generates, fill in much of the volume increase that we were partially attributing to compression or 'instant compression'. When adding every variation of real tape samples to TB+, adding R2R machines to TB+, and comparing these to actual tape machines driven, and finding that they mostly nulled out even the overtones, what was left at many different levels were the variance in what was being compressed. I figured this would be fairly significant, but not really. TB+ was doing a LOT more of the effect of low level leveling or limiting than expected. This gets incredibly complex so I have to cut it short, but the answer is that now with R2R and TB+ not only do we have the sound of the specific machines, but we also add nearly as much drive and harmonic saturation as we wish, and when placed on top of the specific R2R machine the result is almost always identical to the sound of that machine when cranking it harder in real life, saturating harder. by layering an additional instance or two of TB+, it layers correctly because Nebula allows it to layer that well. So, some of the answers about what exactly makes actual higher nonlinear drive and tape compression were answered in this process.
The 3rd stage has always been a part of the concept in completing the last 5-7% of the sonic equation, making it easier to take any of the R2R machines all the way to the max without losing the realism, without undoing something that TB+ already does perfectly, and without sounding like a plug-in. It is too early for me to answer any specific questions about this stage but I am preparing the details and faq's already, and almost everything you may be wondering about this stage will be answered in the audio samples I've designed. The third stage does exist. It will make you very proud of your R2R machines and how well they have accurately maintained the concept as it began back in 2005. How does the 3rd stage affect R2R and TB+? That is a very detailed and even longer question to answer and it isn't yet time for that. I can say that without going through the process of making R2R work, I don't think that anything based solely on any kind of algorithm could have been possible. There have been good efforts from several other developers for several years now, but everything that I've heard has had a sound, and that sound has usually been because of the code and not because of the sound of the machine being emulated.
The 3rd stage, like any good digital audio tool, will add to and enhance what is already working correctly. It does not replace something else that you are already using inside Nebula, but has been extremely carefully crafted to work with the machines created. I think you'll hear that the audio speaks for itself when the time is right. It does not take the place of R2R or TB+, rather it shows how important the effect of the details of the actual real machine affect anything algorithmic that may follow. I believe this is one reason that tape saturators and tape compressor plug-in emulations have always come out sounding like plug-ins. The other element that I'll mention briefly is that something very very subtle is happening in every machine I've tested at any level. To some degree, there is a measure of compression happening all the time. This can be almost undetectible up to a point, and then when we hit the levels hard, we get more and more drive and compression within increasingly smaller integers. When it gets extreme, this is where all of the code-based plug-ins seem to sound wrong to me. I think you will find that by taking an entirely new approach, that sometime in the near future, it will be the first time that a new coded plug-in concept will not disappoint in this area. In absolutely every scenario from doubling, tripling, instances, it never stops acting like tape at more extreme settings. when used after R2R and TB+, you may find very little need for separate mastering comps, mastering limiters, etc., or even better, the ones that you already love using on your master buss will work even better with even less volume needed. So, yes it is confirmed, but please wait for other answers. I promise to share new info as soon as it is the proper time. Really! I promise! I will tell you more soon!
thank you for the detailed reply (as always!) Thats a fascinating read! I'll have to go over it again when i have more time to absorb it all. Actually i am VERY excited to hear this 3rd stage is in development! Any idea of an ETA Michael?
Thanks! You are welcome. ETA- as soon as we feel confident that we can guarantee the same smooth process for each crazy PC OS we will choose the ETA. The sound engine is complete.
sfunk wrote:thank you for the detailed reply (as always!) Thats a fascinating read! I'll have to go over it again when i have more time to absorb it all. Actually i am VERY excited to hear this 3rd stage is in development! Any idea of an ETA Michael?