Roman

yes, that does sound a bit like "rocket science". I, and I'm sure many Logic Users, would certainly appreciate having a Layer like that.

Get well soon [:)]

Nigel

@Nigel Watson said:

Roman

yes, that does sound a bit like "rocket science".

I originally studied to go to the US Air Force Academy. Rocket science is much easier.

[H]

Hello again,

as I have been experimenting with the rep.-tool (don't we have a "republican-tool" to switch some of those guys off in time, hehe) recently - I even read the - btw. excellent - manual carefully this time, line by line [[;)]] - I just wanted to tell you what I am thinking about it after all. To make this post not too lengthy, I try to nail my points on the board one by one:

1. I understand now, how it is possible to create very convincing repeated note's lines with various dynamics and/or tempi. The tool offers many more ways to manage repetitions than anyone was able to check out in a short period. I know I will love the feature in quite some occasions being in production. I consider it to be a fantastic toolbox for digging into patterns and rythmic/minimal stuff.

2. Nevertheless I still don't think, that the achieveable results justify the repetition-tool's long learning curve/awful handling and the time-consuming setup. It feels like filling the bathtub with a liner-cap, somehow. The tool offers endless repetition-patterns, only that I don't find it essential to have them. Yes, it is the LATER thing currently, but this section cries for the NOW-approach: Let's have the NOW version LATER, OK, Peter?. I don't see any reason, why the UI shouldn't get much more simplified in the future and I miss one IMO very important factor: randomization (of note-cycles between start and end-note AND of timbre as well as - optionally - tuning to mask the cycles).

3. Having to deal with note-tails at the BEGINNING of each repeated note makes it very difficult to PLAY repetitions, the delay-compensation for the start note seems no big helper, if you don't substitute pre-recorded lines by a repetition-performance. Timing is hard to get right. Furtheron, the concept limits the tempo-range where repetitions sound good and notes connect seamlessly dramatically.
I experimented with staccato brass-samples, that have natural decay. As soon as I have enough samples to alternate after an accented (short) starting note, the note tails don't run into phase interference while overlapping each other (using random that avoids direct repetitions: 124532154213415 etc.). When I add random alterations of brightness (slight) and tune (+/- 5 cent), the sample-repetitions are hard to get recognized as such. When I place a note, that is a little longer than all others and a bit more pronounced at the end of a repetiton-phrase, the whole thing sounds perfect to me. The achievable "good" tempo-range is much larger due to the fact, that there are no tail-interuptions by design. The timing is easy to nail down. I don't see why the repetitions shouldn't get automated to get widely playable AND post-fiddleable repetitions at the same time - e.g. tapping the pedal for the last note (I love to have my left hand free for switching articulations via keyswitches). Sorry if I repeat myself (a real repetition-post, eh?).

4. The cresc./dim. repetitions sound fantastic, if I use them as recorded. But of course they can't get automated in a "similar-to-3." way, because no tool will ever be able to anticipate the number of repetitions you are going to play to adjust the cresc./dim. "curve" accordingly. I have to repeat myself again: IMO nothing beats the mod-wheel-to-dynamics (xfade-layers) concept. As long as we have enough dynamic-layers (at least 4 for best results), I prefer realtime-control and freedom of curvature to depth in natural timbre. With a patch/tool-combination as described in 3. , that was constructed as a multi-layer-xfade-patch, I had the choice to play whatever dynamic progression within ANY repetition-line SPONTANEOUSLY. OK, maybe I am one of a rare species, that still plays and improvises lines, but that is how I see it. Besides - Peter mentioned that Perf.-Legato is the star in hollywood - I am convinced that such a NOW-repetition-thing would

SELL LIKE HELL (in case there is a chance to top your sales at all [[;)]]... )

Instant gratification. That is where it goes to - everywhere. As it seems to me. At least.

And please note, that I appreciate very much to have the repetition-tool even as it currently IS. I just can't shut up as long as I see the potential to go for more. I am thankful, that this tool has made me really think about the whole repetition-issue at all. Lots of manpower in there, I have deepest respect for what VSL comes up with NOW. So many fantastic ideas at your fingertips, that of course I am nitpicking somehow. - Just to be understood right.

All the best


Roman Beilharz

PC & Musik, Germany

Roman, I have VSL perf. for about a week now and could not agree more.

Roman,

Caroline and I both read your post and we do agree with you. Caroline also noticed the timing issues with those samples. Unfortunately, the last version of the manual we did was the day the Performance tool was released. So as it was being released morning time in Vienna, 3AM in LA Caroline was finishing it and sending it over! So we had no "play" time.

Overall, not to be critical of Herb or Chris, we did feel that the Repetition tool was not NOW oriented enough. In fact, in all our sales presentations done in our studio today, we don't show it, because once composers saw it in earlier demonstrations, the immediate reaction was, "Oh my God, what's THAT!?"

The next reaction was, "Holy S***! How long does it take to learn that?"

Thus, we keep the focus on the Legato tool and the Layers because that gets everyone going the quickest.

You know, we have a 35-lesson Logic course. And as you know by experience, Logic is a deep program, and many have difficulty getting going with it. I can't begin to count how many hours I put in to make Logic so simple that within 6 hours, someone could be sequencing with it and feeling good about the experience.

I feel the same is true with the Repetition tool and to some degree with the Alternation tool. But it takes time and organization to know how to make it simple so the customer can be using it NOW. For example, I'm reading through the strings "manual" now. Nearly 100 pages! It takes a lot of reading to find the articulations that can be applied with the Alternation tool. And the way the Instruments are named doesn't help. I can't just look at a VSL Instrument's name and know/guess what it is. I have to look it up every time.

That's time consuming and it adds to the learning curve for the other tools.

I have the same critique with GOS, SISS, and the DD naming structures. And when every company has its own naming convention, it's a nightmare to figure them out unless you keep a notebook.

So it's not just the UI, it's the naming conventions, too. And there's one other issue: orchestration knowledge.

If the user doesn't know his/her orchestration (and a lot DON'T), then the tool's are even more complex.

So when you add it all up, these tools require:

1. a clear naming convention for the instruments
2. a clear actionable user interface that looks simple and inviting
3. a practical orchestration knowledge
4. a practical knowledge of sound reproduction to understand tails, and other terms.

All of these design issue affect the NOW usage.

But in saying this, understand I speak for myself and no one at VSL.

PA

To check where you can use the alternation tool is not so difficult.
Genreally it works for mappings which are using key switches.

In the orchestral cube you find all keyswitch combinations under the Label name "COMBINATIONS".

If you see at the end of the intsrument name "1+2", you know that there are two variations. If you see "all" you have more, and so on.
If you analyse one string, woodwind and brass instrument, you should be able to manage each instrument and the labeling.

Further the Alternation Tool works for all BASIC INSTRUMENTS.

You can also combine very easy runs and 0 runs. (Performance Set).
(I'll post an example in the upcoming midifiles)

And at last, all of you, who are not afraid of the giga editor, can build their own combinations.


I would say, the difference between legato mode and alternation/repetition is:

In legatomode you simply play.
For the other tools (if you want to use them advanced) you should have a concept before playing.
If you simply use the alternation tool to switch between two variations, of course you don't have to plan very much.

Herb

@herb said:

If you analyse one string, woodwind and brass instrument, you should be able to manage each instrument and the labeling.

I think the point Roman is making, Herb, is that this should be laid out so that the user doesn't have to analyze. It's there before him, so that he can quickly go to the Tool and use it fast.

Although I am aware, that this discussion leads far beyond its original intention, I like to add some thoughts I have about controlling instrumental samples and audio workstations in general. Peter you pushed my buttons again, thank you for your additions and explanations.

Logic Audio is a good example that I know very well (I guess that's why you choose that comparison, hm? I see you are a very skilled pedagogue, Peter! [:)] ). I am really into that program, because I have been learning computerized production of music using "C-Lab Supertrack", the grand-grandfather of Logic, on a good ol' C-64 when I was around 10 years of age. My cousin had one of the first MIDI-based studios on the planet and I had the opportunity to learn the tools right from the fall of men - witnessing the labor pains of this way of producing music. I had been recording my first dubbed tapes when I was 6 years old. I recorded a beat on my drum on one cassette and recorded its playback while plaing the guitar along with it using a second cassette recorder. Then I added the piano by playing the second generation and recording a third. More didn't work well, because the quality degraded too much. As I entered the computerized section, I realized things would probably become more comfortable in a very exiting way. Then the journey took me to Scoretrack, then Creator 1.0, Notator on the Atari and finally to Logic and then Logic Audio on the Mac. Now I am on the PC and quite happy to be stuck with 5.5.1. My Emagic-journey has ended somehow [[;)]], and I don't even regret it. Maybe I will go for the Mac again someday, when they decide to build a true Pro-Line again. I hate small cases with only 2 free slots that I can't configure in a way I want...

Sorry, back to the topic: Although I really know how to get the best and almost all I want from Logic (as long as I have ReCycle and Acid as helpers), I think it is seen from the perspective of handling and ergonomic structure the worst program after Samplitude 5.x. Besides their quite complex specific basic-structure etc., I think in general, that there are much better ways to control parameters on a screen than by simulating knobs and pot's having to "turn" round. It is like having to type a text on a screen-keyboard by clicking the letters. Controllers that simulate the original physic of a mixer can be a helper for us crusted relics of the analog days, but to me it looks like "two steps ahead, three steps back". Back to the future - welcome to the eighties! Only that you can automate everything - not that I am not appreciating this. It is just a rather conservative approach, despite all the hype about the admittedly amazing possibilities that we have now. Imagine you bought a Synclavier 15 years ago and you get warped to 2003 where a guy buys a discount PC for $800 and blows your head off using a hand full of app's... I know all that, but time runs fast and I have to say: Where are the truly visionary concepts amongst all those hopelessly familiar looking hardware simulations? Melodyne maybe. Its key-features should be a basic part of upcoming sequencer-systems. Freedom of time and pitch right at the timeline. But you know, all those funky screen mixers... - And the upcoming generation of musicians/engineers maybe only knows a real mixer desk from the museum - forget they have had hands on or even routine with a dinosaur like that, so why should we go on cloning questionable and limited-to-the-means ergonomic concepts, when the technology was able to offer macros and voltage-free-processing that would overcome the limitations of analog audio and the physics of modifyable resistors easily:

Imagine I recorded a bass, the system recognized the instrument as such and offered me pre-defined parameters/modeling-functions for a bass to mix by default: Fat/Thin (intelligent and signal-dependent EQ/BassEnhancement, Soft/Tight (intelligent Enveloper and Compression), just to mention two ideas. To achieve the intended result (let's say a fat sound that has a soft attack) I painted a fat line with my finger that looked like a roll-off-symbol on the touch screen of my tablet: Such a thing would just feel natural for a homo sapiens: ergonomic, single-surface control (looking at the screen while fiddling at a separate controller to me is a double-surface-task which is much more abstract to do). With a single wipe of the finger we had been substituting around 10-20 mouse/keyboard-actions (open EQ, adjust Q, Gain and Frequency of 3 Bands, open Bass-Enhancer, adjust 3-4 parameters, open Compressor and so on). Of course it was necessary to have all the detail if you liked, but not at the first sight!

Thinking about instrumental-samples my "soon"-vision looks like this:
I have about four layers of looped performance-legatos, that can be morphed in realtime using the mod.-wheel (layering always multiplies the voices needed). Release triggers allow to stop the notes at any time with a natural decay. When I turn up the PBend wheel, the sound would increase to vibrate. When I turn down the PBend wheel, the tonal content would decrease and noise would increase (bow-noise for strings, breath/air for winds). A macro-keyswitch like C1 would make me play those brothers. D1 would give me the same options with grace legatos, E1 would give me short notes, that can get repeated naturally by just playing, an end note would be tapped (as before). Release triggers for these short notes would allow me to play very short and a little longer short notes within the same patch. More macro-keyswitches would give us more special articulatins (pizzicati, tremolo etc.). We could go from any articulation to the perf-legati, the tool handled the necessary "ghost"-key for the natural transitions between articulationsautomatically.

The alternation-tool btw. is a great automation to organize all set up macro-keyswitches - now and then. All I moaned about referred to the repetition concept only, just to make sure. With an improved GUI, it will be more easy to preset figures and ornaments like triplets etc.. I would just name it differently - how about "Phraser" [[;)]] ?

Ooooops, this really lead somewhere. I hope noone minds. Now I REALLY have to write more for the mag and less over here! Holy sh**! Excuse if I went too far, but the whole thing really made me glow...

Roman

Well, I want to say one thing else that I hope you consider about VSL for PC & Musik. Whatever "critiques" we have of this very early version of the Performance Tool, and specifically the Repetition tool, whether NOW or LATER, it IS.

Herb did it!

For everyone who wished you could do what the Perf tool does, now you can. It's new. It's different. It's slightly complex right now. But nonetheless, it IS.

And that's worth celebrating for all your readers. Because now they can do what before they couldn't. And the best, and easiest, is about to come.

PA

Peter,

be sure, that I will of course focus on the amazing things that are possible right NOW with this doubtless milestone in orchestral-sampling. I will encourage everyone willing and able to use orchestral samples to support this large-scale and hopefully long-term-project by buying the First Edition-package. Then and probably only then we might see our wildest VSL-dreams come true someday. On the other hand I have to tell the readers, that what they will get is not yet the final word on orchestral sampling, it is the start - something like: "On the first day, Herb created the first edition..." [:)]. I am already a VSL-addicted, though this fact might have get lost between the lines. I can only hope, that the VSL team is still curious and breathless enough to take my suggestions into serious consideration for future developments and doesn't feel rejected by me pointing out what I consider to be worth improving quite a bit.

Love, Peace and creative visions that make people sigh in adorance

Roman

@Roman Beilharz said:

"On the first day, Herb created the first edition..."

Yeah, BUT!

Now, I'm not speaking officially for anyone except for myself. But I'm not sure that's quite fair for a couple of reasons.

1. I've been involved with this library, albeit on the perimeter, since last July. Originally, this library would have been in Giga 3.0. However, no Giga 3.0. So do you wait or do you go with Giga 2.5? Well, like Doug Rogers at East West, Herb also had a tough decision. And they chose, and I think wisely, to roll out with a 2.5 edition that's 16-bit. That was a business, not an artistic decision. Had there been a 3.0 out when promised, there would be no, "On the first day..." The failure to deliver 3.0 was a major blow.

So, had Giga 3.0 been out, you would have seen a slightly different library in 24-bit. So I think you have to give VSL a lot of points because of Tascam's miss.

2. Unlike any other library, this library is ongoing. All other libraries have a session that begins and ends. But VSL, is continuously recording and continuously expanding. The Orchestral Cube/Performances was supposed to have been Box 1, followed by Box 2, followed by Box 3, etc. In other words, the original design called for VSL to be growing starting with a basic set, and then add-ons. In our July meeting, we were told that when recording was completed at the end of 2003, we would be looking at 2-4 terabyte library.

In all fairness to Herb and his investor, I think you have to deal with the problem squarely by pointing out the business direction shift because Nemesys did not deliver as promised.

Every major developer will have a Kontakt library out by summer, either as a reprogrammed library or a plug-in.

Yes, speaking as an outsider for VSL, I do think your ideas have great merit. BUT! They must be tempered within the current business (not artistic) climate.

So consider these points. Again, I'm not a VSL employee. I'm not paid by them to say this. But I think these are important points that the readers of PC & Musik should also be aware of. It's not just art and tech specs!

BTW, I've moved over to Cubase SX from Logic 5.5 on the PC and am I glad I did. Write me and I'll tell you about it.

Peter Alexander
lacomposers@msn.com

Thank you Peter,

I didn't know about these things in such an explicit way. I myself never have had high hopes, that Giga 3.0 would be out in time. Of course now it looks like 3.0 will be more than late. I understand your point and I appreciate the decision, VSL has made, very much:
To me a 16Bit-library is just perfect, because I consider the desire to deal with 24Bit-samples to be mere hype. As for a sample-library you can even normalize the pp-samples before dithering/downsampling (attenuation after-the-fact brings the pp down to where it should be) and since maxed out and well-dithered 16Bit sounds identically to 24 Bit for a homo sapiens, I don't see any reason for doing a 24Bit version of whatever library at the moment. The only thing that happened in 24Bit were vanishing system-resources and worse performance/polyphony. This is the part that still runs behind all those amazingly large samples: the capability of processing enough of them simultaneously. I know: the 24Bit is better to get post-processed and mixed because of less degradation while again expanding to 32Bit float/redithering etc.. But for me this more theory-oriented point (I can't hear it...) is much less important than being able to BETTER CREATE MUSIC (and this is connected to low system-resources/bit-resoltion again etc.). The VSL-library as it is NOW seems to be the best proof, that 16bit can sound just OVERWHELMING.

Besides all that, you are right: I forget the "real-life"-part of it all sometimes when it comes to optimizing ergonomics. I am a freak.

But my intention has never been to go like: "Bad ideas over here..." - the opposite is true. I ever wanted to give hopefully valuable input, to give some of my ideas back to the folks here - and them. See it as a sign of appreciation for the endless and passionate effort herb and the team have been giving to us all. For the sake of incredibly realistic orchestral samples.
And they are there - already.

All the best

Roman

Roman,

That's the first time I've heard anyone say they didn't really want 24 bit samples - I also don't because there is very little audible difference depending on the quality of the mix and effects, etc. But there is a HUGE difference in quantity of data, which means that any system is going to be taxed even more - and it is at maximum right now on just about any computer trying to shoulder the burden of Gigastudio and VSL. I would personally like to have all the instruments of the pro edition in 16 bit, though I'm sure that statement will generate a collective disgust for my filthy low standards.

Sincerely, William

@William said:

Roman,

I'm sure that statement will generate a collective disgust for my filthy low standards.
Sincerely, William

NO...., NOT TO ME.- To me , this makes sense!

Iwan

It makes a lot of sense to me too!

Martin

IMO it depends on the recording quality in the first place, the instrument choices, AND "collective" properties.

One cannot just compare single samples alone, you need to compare complete mixes/pieces. This may be a key issue. The difference could be minimal, but the subtleties can be something that affects us on a subcomcious level. Obviously, music is the most important key, and if its music for some other form of media other than listening its even less of an issue since its hidden behind other stuff....but "better" is usually better.

Also, if these samples are recorded in 24 bit in the first place, then I'd rather have them in their native recorded format. No matter WHAT the dithering algorithm is, it can affect the sound, and is probably most noticable in the softest samples.

Actually, you guys askign for 16 bit are probably in the majority, You're thinking about System taxing, which is the first thing most software sample users think about (no matter how many people are wondering why Giga doesn't support 24 bit), you aren't thinking from an engineering aspect. We, who are asking for the 24 bit are beginning to look like Audiophiles with all the discussions about machine/system performance.

i personally dont get into this discussion much, because it gets heated. In the end I dont care. I've got a lot more to work on in music than to worry about what kidn of performance hit this will take AND sound quality issues.

I completely understand the concerns in respect to system performance (coming from the Dark Ages when sampling libraries where judged by their smallness - given the 1 or 2 MB RAM hardware-samplers offered 12 or 15 years ago --- but I degress :-] ...) ... uhm - while I completely understand this point of view, you all will admit that a project like the Vienna Symphonic Library can't rely on yesterday's standards, to begin with.

What we are after is to get as close as possible to the technical boundaries of today, or even better: of tomorrow. That's why we not only want to release in 24 bit, but also in 96 kHz as soon as possible. Sounds frightening, maybe, but rest assured, computers won't get slower in the future ;-].

And on a sidenote: actually, all our samples are stored in 32 bit-floating format on our servers, to ensure maximum data integrity throughout the whole editing-process with all its many steps.

That said, I should add that the downsampling algorithm we had developped especially for the 16 bit / 44.1 kHz release comes from a small Swedish team of over-the-top DSP programmers you will hear more about in the not so distant future, I'd say :-] ... unsurpassed quality was our key-issue here, too.

[/END of commercial plug] [6]

/Dietz - Vienna Symphonic Library

While I understand completely, that a "Pro"-Edition has to go for 24/96 regardless of even worse performance of a technology that lags way behind those trends in terms of having to cope with a sample-monster like VSL, let me just vote here for an upgrade-deal for all First-Edition users, that would give them all the additional instruments at 16Bit. Wasn't it a matter exchanging the wave-pools of identical .art-constructions?? This shouldn't be too much of work, I guess?!

Or maybe you would like to poll for such a thing?

Interesting discussion, I think, although beeing way out of it's original intention.

Roman

@William said:

Roman,

But there is a HUGE difference in quantity of data, which means that any system is going to be taxed even more - and it is at maximum right now on just about any computer trying to shoulder the burden of Gigastudio and VSL.

An increased performance burden with 24-bit samples may seem to be true but may very well not be. Since modern PCs are operating on 32 bit boundaries (i.e. the machine instructions work with 32 bit words), the increase in sample bit depth to 24 or even 32, should have no effect on performance. As a matter of fact, if you design your software to specifically work on 16 bit boundaries, and not 32, you end of adding in overhead since it takes more instructions to deal 16 bits then 32. This is why, among other good audio reasons, GigaStudio works with 32 bit data after it has read the 16 bit data in from disk.

Soon, processors will have an internal word size of 64 bits and will then process samples on 64 bit boundaries.

I have also wondered about the benefits of 24-bit samples and expect there are psycho-acoustic studies that both quantify and qualify the difference. I also wonder about working with 24 bit samples in a 32 bit environment.

Craig Duke

I suspect that the performance hit incurred by 24-bit samples isn't going to be too significant. My real concern is the hit to RAM. Someone correct me if I'm wrong, but the same instrument loaded in Gigastudio that uses 24-bit audio is going to use more RAM than it's 16-bit counterpart. If I can only load 75% as many instruments when the samples go 24-bit, that, to me, is not worth the price.

However, if Giga 3.0 comes out in a few months, and magically allows us to use our motherboards that support 4 gigs of RAM, I will happily withdraw my objection.

I wonder how big a priority it is for the T*scam people to allow more RAM usage. I worry that they may be spending all their time on bells-and-whistles like new giga-specific plugin formats and the like. I, of course, have no idea.