Synful Orchestra - wow!

[Yossarian]

A quote from the developer of Synful Orchestra on the Northernsounds forum:

Hi all,

I’ve worked on Synful Orchestra for about eight years so it’s very gratifying to see all this discussion.

My experience with samplers goes back to around 1983 and my first large sampler design was for a company called Waveframe in 1986. At that time I was already frustrated with the limitations of samplers. In 1987-88 I experimented for a while with physical modeling (some of you might know this type of technology as waveguides) but I abandoned that approach in 1988. I came back to the problem of music synthesis around 1996-1997 with a determination to do something about what I saw as the limitations of samplers and sample libraries. Synful Orchestra is the result.

Synful Orchestra is not a sample library.
Synful Orchestra is not a physical modeling synthesizer.

I call the Synful technology Reconstructive Phrase Modeling (RPM). This is not a hype marketing term. Synful Orchestra with RPM really is new and different in the way it generates sounds. There is a page on www.synful.com on RPM technology that discusses how Synful Orchestra works. There is more detail in the Users Manual and for the truly insane there are the three patents referenced on the web site.

I want to be able to play natural instrument sounds from a MIDI keyboard (or other controller or sequencer) using standard controls (Volume, Mod, Pitch-Wheel, Velocity) in a convincing expressive manner. I want this to be a musical experience without a lot of editing, pushing of special buttons, program changes, pushing special keys, applying special phrasing tools, etc (programming as we call it). When I play legato I want it to sound legato as it would on the instrument I’m simulating When I play detached I want it to sound detached.

There are several reasons this is not possible with samplers (no matter what the library).

1) The samplers and libraries do not have the variety of note articulations and especially note transitions reflected in real instruments. (I don’t think making such a library would be practical but I’ll discuss that another time).

2) The samplers and libraries are not able to sustain notes for an indefinite length of time, looping does not sound convincing.

3) One cannot change the character of a note in a natural and convincing way while the note is being sustained.

4) There is not a continuum of timbre from loud to soft as in a real instrument – we have 2-4 “velocity levels” – but what happens when the loudness changes during sustain?

5) The samplers and libraries do not have convincing control over vibrato so controlling vibrato onset in an expressive manner is unconvincing.

6) The samplers and libraries have no natural strategy for adjusting articulation according to the players phrasing. Your obliged to use a bunch of supplementary meta commands to select between a very limited set of articulation alternatives (legato and non-legato) and dynamics.

I believe the last point could be improved somewhat (although not sufficiently) with smarter samplers but the first 5 points cannot be addressed by current sampler technology because they require a more flexible representation of sound than traditional PCM sampling.

In Synful Orchestra sounds are not represented in memory as a string of PCM samples as they are in a sampler. They are represented as time-varying sine waves (harmonics) plus noise elements (attack and transition noises and sustain noises – bow and breath).

This “additive parametric” representation has several advantages:

1) You can change the pitch of sounds without altering temporal envelopes such as the shape of an attack or note transition, the rate of vibrato etc.
2) You can splice sounds in a very smooth fashion without getting a click or the kind of wah that occurs with a cross-fade (which is due to phase cancellation).

In order to get a wide variety of phrasing and articulation, and note transitions Synful does a lot of splicing and pitch shifting of elements from its RPM phrase database and this splicing doesn’t always occur at the beginning of notes. The additive representation allows us to do that in a smooth way. The traditional sampler PCM format does not.

Synful Orchestration leverages the parametric additive representation to sustain notes in an intelligent “statistically sophisticated” manner. No loops. This would be impossible using a PCM representation.

With the parametric additive representation Synful Orchestration changes the timbre when the Volume pedal changes so you can create convincing crescendi and diminuendi. This would not be possible with traditional PCM sampling.

In Synful Orchestra the mod wheel controls vibrato in an “organic” manner. Real vibrato affects all the harmonics of a sound in a complex way (it’s not just pitch and volume) so if your going to increase or decrease vibrato with the Mod wheel you need to have control over the harmonics.

Synful orchestra has a complex algorithm for selecting what note transition or piece of note sustain or attack etc (what I call phrase fragments) to use in a given context as a function of input MIDI controls. This is very much more complicated then selected a p-mf-ff legato or non-legato sample. But this system wouldn’t work unless Synful could “morph” the selected fragment in pitch and duration and timbre to fit the context. This morphing is not possible with the traditional PCM sampling representation.

There are other issues that Synful addresses (like natural portamento) but you can look at the docs or the website to find out about this.

Bye for now,

Eric Lindemann

[René]

Thanks Yoss. Really exciting thoughts.

-René

[Yossarian]

How are you getting along with the trial version, René? I like the strings but I can't seem to get anything like that mp3 demo out of them. It's very responsive to phrasing, both velocity and the timing of the transitions between notes - I guess I just need to put some more time into it. And I have no idea how to get the glides I can hear in the demo.

Someone reported they had limited success when loading some pre-recorded midi files into Synful Orchestra. I can believe that - this is a different beast to a sample library and it has to be played differently.

/Yoss

[Timfonie]

Very impressive, very innovative!

B.t.w. did you notice the impressive curriculum vitae of Eric Lindemann as well?
He is both a composer & sound designer AND an engineer on the hard- and software side of digital signal processing.

"Designs advanced signal processing CPU" for Cirrus/Logic. -> That's not something anyone can do (understatement).

Nice to know about this gifted man.

[René]

Pretty well. I'm still learning how to play it though. Took me a while to figure how to use the modwheel and volume pedal, but it's running well now.

The thing excels on legato, note transitions, vibrato depth control and seamless transitions from piano to forte, with the correct timbrical changes. For some reason, I can't make the Synful Pitch Wheel to work yet, possibly my lousy config here (x-mas at parents). I'm getting good results with the violin, clarinet and oboe, beyond expectations really.

I took a peek at the phrase articulation by opening the .isr in Audition, very interesting. I wonder what would happen if the articulated interval is not in the database, I think I'll email the dev.

In any case, this is definitely not a MIDI Jukebox machine. It requires specific MIDI arrangements to sound as intended. For instance, two notes will play legato if they overlap for a while but not much, otherwise it'll play a chord. As is, it offers much more playability than sample libraries. It's very common to end with 8 articulation files to play a 10-note phrase with VSL.

In any case, I happen to love the violin. It has some breathing that I'd like out and vibrato speed control, but this offers a new tool I didn't have till yesterday, and a definitely new synthesis model which is worthy to look into.

-René

[Timfonie]

I think it sounds a bit dull, and some phrases sounds bad, maybe is my poor knowledge with this plugin.

If that's a flaw of the synful orchestra it can only get better in future versions.

Speaking about the future, taken from synful's website:

Future Synful products will focus on jazz, R+B, rock instruments and roll-your-own new instrument creation.

.

The future is gonna be ever more expressive!

[-SPYRO-]

Yes! absolutely, imagine this on plucked instruments

[-SPYRO-]

roll-your-own new instrument creation

[Yossarian]

roll-your-own new instrument creation

Agreed - will definitely keep my eye on how this develops.

A Northernsounds member has posted a couple of mp3:s made with the strings from Synful Orchestra in their forum:


www.mimesc.com/music/syntest.mp3


www.mimesc.com/music/SynfulSoloViolin.mp3


/Yoss

[chagzuki]

Yes, and they are quite amazing for what they are: apparently unnedited midi files from pre-existing compositions.

[YBaCuO]

Also here:

http://www.acidplanet.com/artist.asp?PID=464974&T=1990

[Joseph Burrell]

I did an AB comparison with the demo midi included with the download with GPO (I did not change the midi, I just changed how GPO responded, ie: set it to change timbre due to velocity and respond to volume in the midi file) and I'm not impressed with Synful at all. The GPO version sounds much better. Period, end of story. Sorry. For what it costs and for what you're getting it just doesn't seem or sound worth it to me. There aren't enough sounds it's half of what Gold or Opus costs for a full Orchestra and double what GPO costs for much more instruments. I just can't see what all the fuss is about. These claims of what it 'easily' can or cannot do are not entirely true and the claims of what the current libraries and samples can and cannot do are also exaggerations. I applaud the technology and I think it is still a very good 'synth' but that's all. For the money, there are much more viable options in the market today. And there are some of them that are just as playable and are actually meant to be played live just like a real instrument. I'm not seeing it. Sorry.

[Lunch Money]

I'm not seeing it. Sorry.

No need to apologize for what you may or may not be understanding.

[Joseph Burrell]

*yawn*

Well help me understand what makes this worth 500 bucks. It has 10 instruments. It sounds like a soundfont to me. The brass isn't impressive at all. I admit that there are some good things here, but the price tag is what makes it hard for me to swallow. I have said, and I'll say it again, it is impressive considering how it is doing what its doing, but is anyone going to throw away their 3 or 5 thousand dollar libraries in favor of this? Not likely. Are the results any better than what can be achieved with good midi programming and any other library on the market? No. Seems simple to me. Drop the price to what I'd consider competitive and I'll weigh it accordingly. With what they're charging for it, I have a right to be skeptical of the claims and I've proven to myself that it really isn't all it seems to be.

[zircon]

The solo violin mp3 posted above is quite amazing, better than anything I've heard - but I don't see how capability-wise this program is really different than say, GPO. Plenty of libraries are capable of changing volume or articulation mid-sustain, for instance. What's the big deal?