I'm wondering how people that comment on sister site comments like:
"This is the best Brass library", "One of the best Woodwind library on the market", etc. came to the conclusion that this is their lifeboat to success.
What it really means?
Let's take for example the latest "hits" Spitfire Woodwinds and Brass. What makes them so special compared to other sampled libraries?

From what I can deduce, they are rather standard sampled libraries in their layout, which makes them unfortunately unsuitable for various tasks. Let me explain.
This is the articulation list for Horn Solo for SF Brass:
A - HORN SOLO:

• a - Horn Solo - Fall
• a - Horn Solo - Long Flutter
• a - Horn Solo - Long
• a - Horn Solo - Rip
• a - Horn Solo - Short Marcato
• a - Horn Solo - Short Staccatissimo
• a - Horn Solo - Short Staccato
• a - Horn Solo - Short Tenuto
• a - Horn Solo - Trill (Major 2nd)
• a - Horn Solo - Trill (Minor 2nd)

I highlighted what I consider not appropriate for modern library to have - separation of articulations based on note length.
So any mildly complicated phrase, which involves combination of notes of various length would be either too time-consuming to sequence, sound plain unnatural, or both.


This is extract from Medal of Honor score, not overly complicated phrasing. Yet reconstructing this using legato/staccato/tenuto patches would be quite a task, more so with the desired doits and expression features.

So what's the solution?

Virtual instrument capable of reconstructing this piece would have to be recorded dry, possibly in anecholic chamber, have plentiful randomization options, should blend seamlessly between volume layers and mimic note start and end based on note-on value or how aggressive input slope from breath controller it receives.
Currently, there's only one match for a "library" that can do the job right - Samplemodeling The Trumpet.
That is what I consider best brass library - as it allows me to phase however I wish to, so I'm not tied to fixed recorded note length, unlike other Kontakt Brass offerings. It's even more powerful with breath controller, which is by the way not supported even on SF Brass.

Those who don't share the opinion, please feel free to step up and provide compelling reasons as to why you would consider SF Brass or any other "regular" sampled lib your best. I would be more than happy to be proven wrong.

 

Last edited: Jul 15, 2017

Yeah this is something what never ends. I also have seen discussions that people are just getting really mad at each other, because they were having different opinions and could not be agree about it with each other.

Anyway in my opinion it's just all about what people find work the best for themself. This is really different for a lot of people and I think there is not really a best library or plugin or whatever. I also think this is great, because music will be much more different than if everyone was using the same thing, right?

This kind of articles are also a little bit useless to me.

http://www.musicradar.com/tuition/tech/the-20-best-daw-software-apps-in-the-world-today-238905

FL Studio is on place one. Probably because most people use this. Does that mean FL Studio is the best one? No, definitely not. Although that article makes it really look like that. Many people see this and think yep that is the best one. I go for that immediately even without comparing the other daws or deciding what the prons and cons are of each daw and later finding out why did I actually go for FL Studio? Please don't get me wrong. I am using it aswell, but I only use it as an example and trying to make my point.

Of course there are libraries or plugins who offers more features and stuff, but I think that doesn't make it the best immediately. They maybe still can be harder to understand. I won't say that's bad though, because practise makes perfect or maybe it just don't really work out for someone. All these little things makes difference.

Of course it makes a nice measure or seeing what other people prefer and hearing their opinions. That's great! I do the same too, while I even said all this. For example I really love Noisia. Their sound design is just on another level and I really like their music. I have taken a look to see what stuff they use and tried to use a few things aswell, but later I feel like yeah this is just not working for me. Although it's not that bad I quess, because it also helps me with experimenting.

By the way time for me to learn some music theory and reading notes so I will be able to fully understand what the images are saying exactly, hahaha.

Maybe not exactly the answer you was looking for, but hopefully you still appreciate it

 

Last edited: Jul 15, 2017

@Nick12 No need to learn notes - this is the extract: www.youtube.com/watch?v=1eLVR22UBYw

The problem is - nearly ALL brass/winds are made this way. And while on some of the recent ones, there's workaround which plays release tail in case the note is depressed before the end of the "short" sample. That's however suboptimal substitute as it creates problems with mushy releases and doesn't always connect timbrally well with the original sample.
While SM doesn't offer exotic instruments in their repertoire - it's still IMHO better choice overall.

SM/OT/Cinesamples Brass makes for a good supporting lib, but not the one that takes the lead.

As for sound preference - yes, some libs sound gorgeous out of the box - but who gets expensive libs just to play sustained notes in standalone Kontakt? Good practice is to road test them and see how they perform in real setting, from which comes my preference.

 

@Andrew You haven't mentioned Kontakt's keyswitching but I assume it's implied since most libraries have that?

What is their method? velocity variations?

 

Keyswitching isn't a solution to this - it's presenting the same problem - prerecorded notes, which you can now switch via keyboard. Useful, but not truly intuitive.

Nope, randomization aspect of SM comes from random detuning and attack shaping - they don't even use RRs. All notes are recorded as "longs", but since there are no reflections to consider, you can slice longs into shorts by simply playing on the keyboard exactly the melody you wish to hear. No need for rigorous note placements and keyswitching.
Although SM does have very useful set of KS too.

 

I totally understand random pitch and it's importance. Not until I started using Jamorigin's MIDI guitar, bass samples had absolutely no life. Once the pitch bend is set to two semitone, vibrato and the randomness of the guitar string's pitch and somehow without hardware the lack of huge latency allows for endless fun and inspiration.

 

For piano libraries it is mostly the feel when I am playing that determines if I like the library or not. If I don't feel like I am playing pianissimo when I am doing so on my keyboard, or have to pound the hell out of my keys to get forte then I will likely delete the library. With so many libraries you really need to almost to go into your DAW and manually tweak all the velocities to get a good sound from the library. 8Dio is particularly bad about this even though people seem to love their libraries. There is also the sound. If I am playing a Steinway sample library I want a warm, buttery, smooth kind of sound from the library not a harsh brittle one. Likewise if I want that modern, bright, almost percussive, pop piano sound I want to be able to find that in a Yamaha C7 library.

For orchestral libraries it is mostly a matter of articulations and dynamic layers. If I can't get a detache passage to sound good with a string library (*cough* CineSamples) then I don't really care how good the legato patches are, the library is unusable in my opinion. The same goes for the natural reverb on the samples. If all the samples were recorded from a decca tree in a concert hall then I am sort of constrained to using only that library for the whole production and it is a pain to get other instruments from other manufacturers to blend well. So many orchestral libraries sound dull and lifeless also and require countless hours of time to get a believable sound from them.

 

@Andrew,

There is another *huge* factor in how a library sounds and that is (drum roll...) the underlying MIDI data. We all know that various CC's control things like expression but if you intend to render audio which originates from a print file (Sibelius etc) be aware that the notes are perfectly quantised and with even the best library on the planet at your fingertips (answers on a postcard) the result will sound like a bad MIDI file.

To get anyway near you near to humanise the data - and there are sooooooo many ways to do that, before you render. As everyone in the trade, any trade, will say; it's all in the preparation!

Good luck!

PS. I love the Spitfire libraries

 

I thought this problem was supposed to be somewhat solved by kontakt scripting?

 

@Andrew

That is IMO the sad sad story behind this big library makers.. they are analyzing the market. And of course, if you look only for the professional, academically trained composers, that's really a small part. So instead concentrating on those, they make people that can't even read a score (let alone harmonize, modulate etc.) believe, that they in fact are composers (in a classical sense).

It's easy, I see the same patterns everywhere: lots of ostinato, pre-made arrangements (by professionals), "new" strange performance/sound design techniques. Long story short, something that sounds "good" even with an octave below and three notes above.

Now, because this is really low quality music, they have to "train" the audience to like it. And they succeeded. Just listen to the music for mediocre series on TV (without watching it). When I'm like in the kitchen and I hear just the TV now and then I literally jump in the living room just to see what's going on.

Only high budget movies get a treatment of "real" composer and a real orchestra. And those who have seen a rehearsal of an orchestra know how much is going on there. First the communication between conductor and the players. And just the look at the score afterwards - almost every bar marked with red, green, blue.. with notes on how to play a certain passage and so on and on.

..so, we want a "library" that comes close to that? I say, forget about it. And we want a library to reproduce a notable composition from the past (no matter what era)? I say, forget about it. I am not a composer. I studied music and piano, but left at some point. But my close relative is a prof. at the academy and teaches the most difficult subjects like harmony, counterpoint, instrument arrangement.. and being close to him, gave me a lot. (Like when he said: "when you know all the math behind music, you still have to have talent!")

Will I ditch these libraries? Well hell no! They are just made for my "mediocrity". If those "composers" are making music for low/mid budget television series, I can do it too (maybe I am talented lol). To be honest, I personally know a good composer that makes those demos (for OT libs i think). I just have to find a way to get into the club. No Andrew, I can't prove you wrong, even if I wish I could..;) The libraries dictate how we write music, not the other way around..

I agree that so called SWAM (Synchronous Wavelength Acoustic Modeling) like you said - especially with breath controller is the most authentic (real-time) equivalent of the real thing, the limitation being the number (and type) of instruments it supports. For the rest of orchestra you will have to "reinvent the wheel" (but this is really a vast subject).

 

Interesting subject :

In my way of thinking the first thing to consider would be that there is NO "Best"

The word itself is often misused or inappropriately inserted as a "label" that describers a users subjective train of thought.
In my own opinion it would be more appropriate often to think about using the word "correct" rather than "best."
In a sense the word "correct" has objective meaning that can be applied to a certain situation rather than a users emotive feelings regarding subject matter.
As for the instruments described themselves?
Well there should be a clear distinction made between virtual instruments and regular sampled libraries. I personally don't have much experience of say orchestrating large classical pieces but can add a few thoughts I guess.
The simple way forward would be to state that in the situation described the user is aiming to pick the correct tool for the job.
If the goal is to play say a certain passage of maybe a solo trumpet with flutter etc then probably Sample Modelling with SWAM engine is most likely to give the user the most freedom and ability to play with a certain authenticity that a traditional sampled library is maybe not going to provide.
If however the users Goal is to recreate say a Horn section\Tuba or etc within the backing of a large orchestra then quite possibly a large orchestral type library such as maybe Spitfire Symphonic Brass or Orchestral Tools etc is going to be the correct tool for the job.
There is of course negative and positive aspects to both that the user must consider.......First size of library is really important
Most people who actually script Kontakt libraries themselves all agree that sampling itself is not the final answer or the way forward.
A 100Gb+ library is maybe all encompassing but as to extending a future then the answer cannot simply be to sample more & more.
"Bigger" size does not mean better and is certainly no marker as for being the correct tool for the job?
A virtual instrument, with maybe a wind controller, small footprint, great expression with huge note choices is often going to be a great choice. Will it be the same for every situation? Certainly not in my own opinion?

Well yes it does to a certain degree but it also has various limitations and cannot be applied to every given situation in a definitive manner. Scripting is limited to a degree and there is only so much you can do.
Aspects such as release notes cause many problems and to put it bluntly there is still many things you cannot achieve with scripting alone.
The fact that it's sampled means results to some degree will be predictable.
Examples would be say round robins or alternations of differing length not being possible within similar instruments. problems with note velocities and midi etc, etc. The list goes on...... Scripting is fantastic & you can do some miraculous things but like most things it has it's limitations.

Personally I don't think there can ever be a definitive answer but would rather state that a user should be objective in looking at choosing the correct tool for the job after considering the various constraints regarding situation & subject matter.
Of course some results involve personal interpretation and it will always be that way as human beings with emotions are involved.....For the particular scenario you describe then i'd most probably agree with your idea that Sample Modeling or a virtual instrument rather than a traditional sampled library is the better choice.

This particular scenario & solution however cannot be applied with any degree of universal exactness

 

Last edited by a moderator: Jul 16, 2017

I usually find the best to be the best. But if that fails, use the second best.

 

That's why synthesizers are still the future.
For complex articulations, just use sample-modeling software instruments, even if that supposes layering dozens of them, in the end, it still sounds better that 200gb fancy sampled libs.

 

Any library which cannot handle basic phrasing, is ultimately not the right tool for the job.
SM manages even bigger pieces with outstanding level of customization:

The above piece would take insane loads of effort on part of the composer to sequence by "regular" means.

Where all-in-one packages excel is simplicity for simple tasks - e.g. 6-horn patch occupies just 1 track and is actually OK for slow/largo parts. And for those situations in which the composer is lazy, as with SM you absolutely HAVE to record unison separately - copypasting MIDI would create problems with phasing.
So ensembles are actually bit more tedious for PM than for sampled content, when easy parts are in question.
This is understandable for strings, as in 60-piece, 30 of them are violins - and recording every one of them might take more time and effort than simply using "1st Violins" patch in CSS.

Scripting isn't solution in real sense - one might be able to set triggers what happens when. It's rarely used to bend the timbre in real time, only to define playback conditions, button functions and positions and keyswitches.

Also SM are quite punctual as to what they call SWAM and what not:

Source: http://samplemodeling.com/en/swam_flutes.php

 

Last edited: Jul 16, 2017

Is it possible at all, to automatically change the articulation based on how long/short the note is?

Also, does this mean my crappy SF2 soundbanks (1 or 2 patches per instrument) with attack control is in a way, better than these big libraries?

 

Good topic. For a terrible split second I thought I read another "best [XXX] lib?". Well, even with this site's name let's say [XYZ] to avoid misunderstadings . And this was the scariest part: started by Andrew (OMG). Luckily I got it right the second time. Coffee helped.

Because some of the things commented here I end up using sample based VSTis rather than Kontakt Libs. My favourite metal electric guitar is Shreddage II, but I end up using Ample Sound's Eclipse 2. Some articulations/sounds are really hard to program in Kontakt, because of the all around sample player design which means not total integration with the effects and specially complex numeric computation (float numbers introduced very recently).

When I said hard it's often also slow. The most powerful kontakt eguitar, Orange Tree's, uses a lot of memory and CPU. Specially CPU.

When I mention effects I'm not talking about the usual reverb/chorus. For example, convolution eq matching plus interpolation is used to simulate different pickups than the one recorded. Who says pickup says capo position. Some things that didn't add up when I first compared sizes of kontakt libs vs Ample Sound and even RealGuitars.

And about numeric computation:

I'm not an expert by far, but I'm pretty sure that these highlighted lines require quite a bit of floating point CPU power. Asides from the proper recorded samples, provbably processed to fit the needs of sample based synth.
Whatever you want to call it; physical modelling or synth: sample based VSTis are the future. Hell, the famous IK MODO Bass I bet it uses some kind of preprocessed samples.

And all I've said is not Kontakt fault. They're just limitations that the powerful features of Kontakt have managed to overcome to very remarkable heights.

As usual, I do everything but music lol. Lately I'm having a crush with basic DSP theory

 

Last edited: Jul 17, 2017

@Andrew

I think Emotive Strings is a good example for this aswell. Not sure, but are these phrases not already totally premade? If it's true, then you don't really have the creativity to write your own things or exactly what you want. That is kinda what you mean, isn't it? I see why people are using libraries from Spitfire actually. Maybe instead of the articulations and stuff, I find the sound quality really good.

By the way, if you want to make this score, do you need to know exactly the used articulations and do you have to use them aswell? But even with a piano you can still write/make this score, right? So you actually would think that this is possible with every orchestral library too, but it seems like it don't. I don't really get it actually, but it's interesting.

This is pretty interesting aswell, because why are some libraries still so expensive if it just really sounds so dull and lifeless? Which libraries are sounding dull and lifeless in your opinion? One library that comes to my mind that I don't like is East West. And this is even a big and a really known one. I just don't really like how it sounds. It sounds so dry to me. Wondering if here are more ones that thinks the same.

 

Here's the Dogs of War excerpt played on SM (transposed to A):


Sounds pretty convincing IMHO , even though it's just a draft - all imperfections are to be related to my playing, not to the shortcomings of the VSTi.

 

I can't play it. Mind posting a link instead?

 

Apologies, this should work better:
http://picosong.com/vn9r/