Prophet X Speculation

[AlanC]

I see a lot of people complaining about the price...but I do wonder perhaps sampling really is an expensive endeavor? A lot of people don't realize that Fairlights were $25,000 and Synclaviers were $100,000. I don't think Dave or Waldorf deliberately set out to make $4k synths but at the end of the day they ended up costing what they did. There must be a reason for it, maybe the processing power or storage of the samples, I don't know. As far as alternatives...well no other synth company now is doing lower priced Emulators or Ensoniq type synths...hell almost no one is doing samplers at all! So until Behringer wants to make an Ensoniq Mirage clone or Emulator clone for $1000...this is it.

Back when the Fairlight and Synclavier came out memory was expensive, and sampling needs plenty of it. Nobody is bothering to build hardware samplers now because general purpose computers have exceeded the capabilities of the old hardware to a massive degree, so you just need software. Case in point: you can get a Fairlight emulation (the full thing) for the iPad for £49.

DSI are offering something different: a synthesizer where some of the oscillators are sample based. Not my personal cup of tea because I have a preference for pure synthesis, whether it be subtractive, FM or additive. That said, I do find some of the Prophet X demos very appealing.

[LoboLives]

I see a lot of people complaining about the price...but I do wonder perhaps sampling really is an expensive endeavor? A lot of people don't realize that Fairlights were $25,000 and Synclaviers were $100,000. I don't think Dave or Waldorf deliberately set out to make $4k synths but at the end of the day they ended up costing what they did. There must be a reason for it, maybe the processing power or storage of the samples, I don't know. As far as alternatives...well no other synth company now is doing lower priced Emulators or Ensoniq type synths...hell almost no one is doing samplers at all! So until Behringer wants to make an Ensoniq Mirage clone or Emulator clone for $1000...this is it.

Back when the Fairlight and Synclavier came out memory was expensive, and sampling needs plenty of it. Nobody is bothering to build hardware samplers now because general purpose computers have exceeded the capabilities of the old hardware to a massive degree, so you just need software. Case in point: you can get a Fairlight emulation (the full thing) for the iPad for £49.

DSI are offering something different: a synthesizer where some of the oscillators are sample based. Not my personal cup of tea because I have a preference for pure synthesis, whether it be subtractive, FM or additive. That said, I do find some of the Prophet X demos very appealing.

This is true. I have the Fairlight App. I was thinking of incorporating some of the sounds to my music.

I think what the Prophet X is doing is taking the ideas of using samples AS oscillators and as such you can do subtractive, FM or addative synthesis but using the sound of a piano or vocal as a foundation which none of the old samplers offered.

[AlanC]

I think what the Prophet X is doing is taking the ideas of using samples AS oscillators and as such you can do subtractive, FM or addative synthesis but using the sound of a piano or vocal as a foundation which none of the old samplers offered.

Some of the older E-Mu samplers did incorporate analog filters and, rather amusingly, they used the SSM 2044 chip, so you could see the Prophet X as a sort of  logical successor to them. Plus the Yamaha SY77/SY99/TG77 had the ability to use samples as modulators within FM algorithms.

So nothing particularly new here, just what you'd expect from DSI: a really nice implementation that makes the whole thing interactive and approachable.

[LoboLives]

I think what the Prophet X is doing is taking the ideas of using samples AS oscillators and as such you can do subtractive, FM or addative synthesis but using the sound of a piano or vocal as a foundation which none of the old samplers offered.

Some of the older E-Mu samplers did incorporate analog filters and, rather amusingly, they used the SSM 2044 chip, so you could see the Prophet X as a sort of  logical successor to them. Plus the Yamaha SY77/SY99/TG77 had the ability to use samples as modulators within FM algorithms.

So nothing particularly new here, just what you'd expect from DSI: a really nice implementation that makes the whole thing interactive and approachable.

Well while the Emulators had analog filters and amplifers they still didn't allow samples to be as heavily manipulated so easily but like you said DSI simply too the concept and made it interactive and approachable.

[Pym]

A few short answers to things I've seen mentioned:

User samples will 100% be in the instrument. This is not like the Tempest where I was clear we were trying as hard as we could but no promises. User samples are already basically working on the synth but we have a lot of other things to get working before implementing, like an easy way for you to make the multi-sample files and import them into the machine. This is easy for our add-on packs because we know how to do it manually, but we want to make it easy for you guys. It will be worth the wait

Slop as a mod source is a slow slewed randomization. Think an LFO that randomly changes period length

The new filter is amazing, the low end sounds particularly good which I think was lacking on our older filter designs. Of course it's all subjective

You can do audio rate modulation of the samples and oscillators to any of our mod destinations. This gets some next level results when you experiment

The sample instruments have 3 different loop modes... pitched, sync'd and normal. Pitched keeps the loop size consistent with a particular note value so it sounds pitched. Sync locks to the tempo. Normal has more granularity. All of them can be modulated

I think technically the hard drive could be replaced by a larger one but it isn't as simple as just formatting it on your PC/OSX computer. This is something I'll revisit after things are released to see what our options are

[LoboLives]

A few short answers to things I've seen mentioned:

User samples will 100% be in the instrument. This is not like the Tempest where I was clear we were trying as hard as we could but no promises. User samples are already basically working on the synth but we have a lot of other things to get working before implementing, like an easy way for you to make the multi-sample files and import them into the machine. This is easy for our add-on packs because we know how to do it manually, but we want to make it easy for you guys. It will be worth the wait

Slop as a mod source is a slow slewed randomization. Think an LFO that randomly changes period length

The new filter is amazing, the low end sounds particularly good which I think was lacking on our older filter designs. Of course it's all subjective

You can do audio rate modulation of the samples and oscillators to any of our mod destinations. This gets some next level results when you experiment

The sample instruments have 3 different loop modes... pitched, sync'd and normal. Pitched keeps the loop size consistent with a particular note value so it sounds pitched. Sync locks to the tempo. Normal has more granularity. All of them can be modulated

I think technically the hard drive could be replaced by a larger one but it isn't as simple as just formatting it on your PC/OSX computer. This is something I'll revisit after things are released to see what our options are

Wonderful news!! Do you have a limit to the length of sample at all? Looking on the front layout, is there a way to change the pitch of the samples at all?

[BobTheDog]

...
You can do audio rate modulation of the samples and oscillators to any of our mod destinations. This gets some next level results when you experiment
...

Hi,

That sounds pretty good, is it just frequency on the samples that can be modulated at this rate and what is the exact rate?

Thanks

Andy

Edit: Just read what you posted a bit more closely, I guess the correct question is what parameters on the sample oscillators  are mod destinations?

[LoboLives]

In the Sonic State Video there is mention of multisampling...would it be possible to essentially create a custom multisample where it has various samples that are split across the keyboard? For instance the first key be a timpani Drum, then the second octave be a bass and then a VS Piano on the next octave and then a string sample on the higher octave? Not with the front layout on the synth but from an external editor or user sample? This I firmly believe would take this thing to the moon and back as essentially you could have a drum sequence on one layer, then a bass sequence on the other (recorded on the lower octave) and then the rest of the keys to play piano or whatever.

[dsetto]

I believe that will happen based on what they've advertised and reassured.

[dsetto]

I've read:
- you will be able to have it so playing successive notes on keyboard changes pitch, or not.
- 1.5GB max for sound loaded in RAM per instrument. Times 2 in one layer. Times 2 layers (split/stack)

[LoboLives]

So essentially it will be multitimbral as you will be able to have more than one sound on a single multisample under a single Instrument? Like every key being a different sample?

[Razmo]

I don't know if this has been mentioned previously as it has been some time since I participated in this thread, but what about the Time Stretch function of this engine? ... is it off-line, or is it a realtime timestretch function?

I'm asking because this is what could potentially do the difference to me wanting this samplesynth... Timestretching makes samples so much more useful because you can get around that typical ROMpler type of sound, especially as loops with rhythmic content can be pitch shifted without the tempo changing as well.

Another thing is that I'm really interested in hearing this time stretching algorithm in action, in a more technical way... how does it sound when stretching? ... how does it work? are there different algorithms for different sample material?

... and also... now that i read it has audio rate modulation of the samples, I think that the link to the P12 code has been cemented... this sounds more and more like a heavily modified P12 engine being yet another telltale sign on why I heard a "p12" when hearing the demos because the audio rate modulation also have a very certain sound to it that I remember clearly from the P12 when I had it...

[dsetto]

Fitting for this speculation thread, here's my summary of reasonably supported heresay. I think most of the stuff I included here are facts. Especially the parts preceded by "I think....". This is my completely unverified grasp of what the Prophet X is. It’s review for most of the regulars here. Please correct misconceptions or errors. (I have yet to see any videos other than Dave's official introduction. And I have never played it nor seen it demonstrated.)

________
NOT YOUR POPPA's MULTISAMPLER
It is not like the user multisampling equipped keyboards from Korg, Kurzweil, Yamaha & Roland. It's got some similarities & differences to the different sampling in Nord.

NOT A SAMPLER
It's not a sampler in that it doesn't sample, and doesn't have direct file-level view or editing of a sample. Has USB port for sample import. On the keyboard, you can load a keymap, but I believe you can't see or edit a sample directly.

SAMPLE SYNTH?
Thinking of it as a "sample synth" makes sense to me. As we know, it has wonderful real-time Sample Controls in that section. Plus, these are in the 16x2 mod-matrix.
________
SAMPLE STRUCTURE
Sample->Keymap->instrument->Layer->Program->Enlightenment

Structure as I understand it:
1 Program can have 2 layers. Split or Stack.
1 layer has 2 oscillators & 2 instruments.
1 instrument loads one "keymap".

Eventually a user will be able to create a keymap on a computer, and load it into the Prophet X.
1 keymap can contain from
- 1 sample, on one key, one velocity level .... to
- Tons of samples, mapped to lots of keys, and lots of velocity levels

A keymap is loaded into RAM. Max per instrument is 1.5GB RAM. Therefore, max keymaps loaded at one time is 6GB. 2 instruments X 2 layers.

Switching programs that have different keymaps loaded into the instrument slots is quick, though not instantaneous. What hasn't been demonstrated is how long it would take to load a Program that has instrument slots filled to their max RAM capacity. I.e., how long does it take to load 6GB of new keymaps into RAM. I have the suspicion it's at an acceptable modern rate. (But, no seamless transition. Not a traditional modern multisample player.)

SAMPLES, ENCASED AS INSTRUMENTS, FOR SYNTHESIS
Key point: The interaction between the samples and synthesis is via the "instrument". Think of instrument as a DSI oscillator. Instrument could be one sample, on one key, one velocity level short or the other multisample, multi-vel-level extreme. And, there are instrument-specific actions that are characteristically useful in manipulating samples. (see Sampler Player controls and related source-destinations.)

REAL-TIME SAMPLE CONTROL
The sample control knobs are the most interesting "drive" controls for the sample side. They include knobs for sample start & stop, loop size and center and buttons for loop on and reverse. Again, these are represented in the mod matrix. It's this classic DSI, heart-of-the matter & synthesis-integrated approach that's an enticing approach to sample playing. (And it's got the two strips on the left.)

________
ROMPLER TODAY, USB-SLOT LOADED USER-MULTISAMPLER SYNTH DECEMBER
As of now, 150GB is reserved for 8DIO library- feeling similar to Read Only Memory sounds. 50GB are reserved for user sampling. There will be more 8DIO libraries available. DSI is aiming for user sampling to become available in December, accompanied by a program to make the keymaps. There are hi-fi, high quality, character multisamples that can sometimes be 10 velocity levels deep. (Intelligent move allowing to leap frog into this game- with its own terms.)

[A week ago- that paragraph in a DSI product wasn't a thing- for most, except a privileged few. We had one here give us enough info to maintain post-teaser, pre-cat-bag sanity.]


SYNTH SIDE
And those multisampling sounds are integrated with the 2 digital oscillator (times 2 layers) synth. Supersaw here. (As in Pro 2, & not 12).

DEEP SYNTH COMPARISONS
The synth side in isolation is a little simpler than the Prophet 12. 2 osc, not 4. Also missing from Prophet 12 are/ L-R wave morphing, character, delay lines, & linear FM. But, there is *plenty* new-to-DSI (& generally exciting) synthesis to be explored because of the integration and interaction between the oscillators and instruments. (That's an understatement; and that is the mystery & intrigue that I imagine has sound hunters revved.) Remember, an instrument is a keymap, and a keymap could be a single sample or it could be a 10-velocity level multisample. The only limit I've seen mentioned is the 1.5GB max loaded per instrument. (Surely, there's got to be a velocity level limit, right? Or, a total sample count limit per playable instrument.)

________
POLYPHONY
Max polyphony is 8 stereo or 16 mono. A stacked Program can have 4 stereo voices, or 8 mono voices. (Not traditional multisample player.) In stereo, a new 24db filter is hardwired L, and another one right. Per stereo voice. A button allows to shift to 16 mono, off of 8 stereo. That's a good idea. Start in one mode, and shift as desired or needed based on the whim of your idea.

EFFECTS
There are 2 effects per layer. (4 total). And, different from Rev2, an additional parameter knob- totaling 3- per effect. (All of which are we represented in mod matrix.) I read that if FX are off, signal is all analog VCF forward. (As in Rev2.)

FILTER
Talk of filter sounding good. It's 24db only. But has a filter drive knob. (Circuit, not discrete, new for DSI.) PEK folk here know real well at what true stereo filter separation means. There's HPF as an effect. On the Rev2 I've found this to be good, with surprisingly no perceptible latency on a one-time check for it.

The instrument (sample) can be variably routed separately from the filter, and it has something like a "tone control".

OTHER DSI COMPARISONS
- 16x2 mod matrix, like the 12, whereas the Rev2 has 8x2.
- Has the polyphonic sequencer of the latest 3 analog synths. Rev2 also has gated "parameter" sequencer.


_________
WRAP UP
It's absolutely a gas to see on the DSI site, to have an offering that reads:  "Great for traditional instruments such as ...." That's new.
Obviously I'm fascinated. I'm not in need. I'll be checking in.


X IS FOR PROPHET
- essential realtime crossfade parameter. This is so important, maybe this is the namesake
- Junction of two worlds: Sampling & Synthesis
- Attitude, supported by the black wood side panels. Nice touch.
- Godzilla vs Monster (?), courtesy of LoboLives

[BobTheDog]

Interesting stuff, are you sure there is only one velocity layer?

Also all samples loaded into RAM rather than streamed?

[dsetto]

I'm not sure of any of it. What I believe is ...

A keymap is assigned & loaded to an instrument slot.

And an 8DIO factory library keymap can consist of many velocity levels.

As few as one (as in possibly the VS wave keymap.) And at least as many as 10 (as that was announced for one sound in the official intro video and I believe some 8DIO copy.)

I assume grossly that a user keymap would have comparable specs.

[dsetto]

Re: sample loading question ...

I believe in this forum a DSI spokesman wrote that an instrument can load a max of 1.5 GB samples into an instrument slot, so they can allow to buffer this times 4, due to the RAM limit.

(Nothing I write is official nor verified. I'm just trying to quickly get a sense of this wonderful instrument.)

[BobTheDog]

Thanks for the info, I guess we will have to wait and see. It will be a shame if they haven't implemented streaming though.

[Hector Space]

As always with these complex instruments, teasing out precisely what they can do and their detailed spec is not straight forward. As the long time owner of a Kurzweil Forte and as an engineer I understand not only the complexity but the necessary compromises. Once you start demanding multi gigabyte samples, things get tough. Firstly the sheer size of data to move is enormous, addressing it directly requires 64bit address buses and the processing to go with it. As we all know Flash Rom can be cheap but it's not fast. Ram and fast N-type Flash are expensive. There are basically two tech solutions - 1, use a solid state disk (like Korg with the Kronos) or 2, fudge cheap Flash (like the Kurzweil Forte) both have draw backs. Nord's solution is to shy away, this is quite understandable. Roland and Yamaha are in a similar position to Nord but have invested more heavily in 'modelling' eg the ghastly Vpiano!
Looks like DSI have chosen the Korg Kronos route.. Solid state disk and ram. This is quite a good compromise once you've got the sample data loaded into ram.. But it will make boot up slow for big sample instruments. The benefits will be the speed of the key-to-sound response, fast ram based play back is always going to be better than the latency you get with Kurzweil's solution.
Kurz's solution gives you multi gigabyte samples - 16gb of which 3.3gb are user loadable all of which is immediately accessible, this is certainly enough to do most things I ever need, in comparison DSI's 150gb x4 of ram is brilliant! The thing with the Kurz though is all samples are immediately accessible once booted. So what this suggests to me is that in reality once you've loaded your sample selection into the Px ram, you're good to go for most things you'll ever need!
Also be warned.. Loading gigabyte samples via USB takes literally hours.. The PX will need USB 3 with the potential data sizes involved.

[Hector Space]

Some more thinking:- I'm intrigued to know what the internal architecture of the PX is. First thoughts are that its inherited the P12's multi Shark DSP approach. I think the P12 has 6 Shark DSP chips, each one provides two voices. Whereas this approach makes good sense for high quality oscillator simulation, it does start to unravel once you add gigabytes of 'common' sample data that needs to be shared across all 16 voices. Basically what this would mean is a massive common address/data bus that is doing DMA (direct memory access) on acid! But funnily enough that's exactly how ( albeit on a 1980’s - 6809 scale!) the original P2000/2002 worked back in the 80’s. Ok SCI used custom LSI to clock the sample data out for each voice, no DSP back then. I had both the P2000 and P2002, great they sounded too with their Curtis filters, even if the LSI chips were flawed and couldn't run at their designed frequency. But anyway.. If DSI has gone the route of reuse and is using a beefed up P12 architecture, then it'll be a very different beast from the Kronos. Even before we get to the analogue filters.

[Quai34]

User samples i the future... where did i hear that before ...

Well there are two options: either buy now or wait until the user sample feature works.

I am absolutely thrilled at the news - but given the cost, I'll def be in the "wait and see" camp. For the price I would have hoped for at least 4-part multitimbral and split/layers. Still, the more laptops this thing keeps off stages, the better! 

I have the exact same idea, I'm still fighting to NOT have any computer on stage and this might be just the answer....