Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards Integra 7

It may not sound right, but that is what the Shannon-Nyquist theorem states. As long as the sampling frequency is at least double the largest frequency you want to represent, the sample uniquely determines the periodic function that was sampled. In fact the definition of aliasing is two different functions that have the same discrete representation after sampling so that the decoding process has no way to know which to generate. The lack of aliasing is thus another way of stating the property that there is a unique periodic function being represented by the particular discrete sample. An anti-aliasing filter removes artifacts that are at frequencies above the Nyquist frequency so that the input meets the preconditions of the Shannon-Nyquist theorem and the representation of that unique input function is represented by the sample, and regenerated during analog rendering.

A sample taken at a sample rate that is at least double the input frequency may be rendered back to analog in a way that differs from the original waveform due to noise from quantization error, which is the error of not having enough resolution in the quantum size to represent the amplitude values of the waveform, and may suffer from clock jitter which happens when the clock does not generate perfectly uniform intervals between samples, leading to some phase shifting in the samples.

These issues, however, are not relevant to the discrete representation of sine waves, square waves, and sawtooth waves, which are uniquely determined by just two discrete values— amplitude and wavelength (or equivalently frequency). If you want to store a sample of discrete values for these, they may be generated mathematically by calculating the values of each function at the points in the desired sample mesh. There is no reason to sample the output of an actual oscillator circuit. Quantization noise will be limited only by quantum size and floating point precision, and jitter only by floating point precision, if at all.

But all of this is irrelevant to evaluating the virtual analog synthesizer in a Jupiter-50/80 or Integra-7. It is a true VA synth engine, and not sample-based sounds. It provides excellent sound quality, which is what matters, not how the sound quality was achieved.

Now I can confirm that oscillators in Integra 7 are sample-based.

If I understood your statement correctly, you are not correct. For a straight line, we still need as many points as a sine or other waves. It doesn't matter if its sine, square or sawtooth. We need the same number of points sampled.

Also, Integra 7 doesn't seem to have a mod-matrix which I am mainly interested in.

Synths like the analog Deepmind would be nice but I want a digital multi-timbral emulator.

The other issue with Integra 7 is the pitch changes. Since it is sample-based, it has the sample pitch change artifacts.

Thanks, everyone for helping me out on this. I can now say that Integra 7 is only a sound module designed to be used as-is and not as a sound design module.

I still appreciate your reccomendations.

I've just looked at the Integra 7 Editor Manual, and I think you're right:

. . . It's a _really_ sophisticated sample-player.

If you look at the user manual for (as an example) the MicroKorg XL+, it has a _lot_ more (virtual) knobs and buttons for you to play with. There's an extensive "mod matrix", inter-oscillator links, several LFO's available . . . You can build _your_ sounds, not just modify those that already exist.

Have you looked at any of the VA VST's ? Multi-timbral VA hardware is hard and expensive to build. Multi-timbral VA software -- that just chews up CPU cycles. So some programmer probably took up the challenge of writing a multi-timbral VA synth emulator.

This is not correct. The Integra-7 is not just a sample player. It is very complex and it would be difficult to understand all of its features just flipping through one of its manuals.

From Keyboard Magazine: https://www.keyboardmag.com/gear/roland-integra-7-synth-module



I would not recommend an Integra-7 if all you want is a VA synthesizer, but it does include a VA synthesizer.

It might not be just a sampler but the SN Synth uses samples as oscillators.

Integra 7's architecture is not complex. Especially the sampler part is very standard and easy to understand. The Synth is also very simplified. The other SN sounds have only fixed set of parameters.

If you look at figure 2 in the keyboard magazine link I posted (which is duplicated below) you will see an image of the iPad app for controlling the VA synthesizer. If you look at the area of the user interface labeled OSC you will see the choices for what the virtual oscillators can generate. You have the option to generate square waves, sine waves, or sawtooth waves, or start with one of 439 samples that are stored in the keyboard and designated for oscillator output. This output is routed into filters so that you can set up a cascading flow of subtractive synthesis to design a sound.

The fact that you have the option of starting with PCM samples does not detract from the fact that you can start with square, sawtooth, or sine waves and build up analog subtractive synthesis flows from them.

https://www.keyboardmag.com/gear/roland-integra-7-synth-module

Fig. 2 doesn't infer that Integra 7 is a virtual analog synth! Fig. 2 exists in any ROMpler.

Everything in Fig.2 has an alternative in MOTIF series (pure sampler). That's why I'm saying it is simple. That's not what I call a VA Synth... may be Roland likes to name/call it, I won't

It is true that many VA synth features are available in sample-based keyboards to modify sounds. The integra-7 supports generating sine, square, and sawtooth waves and modifying them with LFO and attack envelope settings, resonance sweeps, ring modulator, and low, high, and band pass filters. A keyboard with o my that feature set would be a VA synth. What you are looking for that the integra-7 lacks other than physical knobs snd other physical controls?

Fixing “spell corrector” error and typo:

It is true that many VA synth features are available in sample-based keyboards to modify sounds. The integra-7 supports generating sine, square, and sawtooth waves and modifying them with LFO and attack envelope settings, resonance sweeps, ring modulator, and low, high, and band pass filters. A keyboard with o my only that feature set would be a VA synth. What you are looking for that the integra-7 lacks other than physical knobs and other physical controls?

Your not going to edit pulse width in a recorded sample. The Integra 7 has digital oscillators, in the synth section. An Supernatural Synth Tone contains up to three partials (oscillators) each with a different wave type including triangle, square, pulse, saw, sine and supersaw plus PCM waves too. The VA synth engine was taken directly from the Jupiter 80 though with less polyphony.

The Integra-7 has the same VA synth engine as a Jupiter-50, which thus has 128-voice polyphony, like the integra-7.

Sweelinck the oscillators are samples. I have sine, square etc. in my MOTIF. It is a sampler. Here is the proof for it:



https://www.gearslutz.com/board/ele...oduction/1179589-future-va-synths-6.html

I'm looking for capabilities such as this:

https://ask.audio/articles/behringer-deepmind-12-the-modulation-matrix-explained

also, a true VA not a fake looped noise oscillator. I have this crap in my MOTIF XF with 8 parts and one common LFO (9 in total).

Integra 7 = sampler

That claim is actually false. In fact, the opposite would be true. If they collected stereo samples of analog renderings of oscillator circuit output, the two channels would not be phase complements of each other. You can produce algorithmically anything you can generate from samples, Nobody claims VA synths have physical oscillator circuits. Virtual analog just means that the components of an analog synthesizer are simulated in the digital domain. If an oscillator algorithm stores precomputed values of a period of a sine function so that it does not have to compute them with floating point arithmetic at runtime that does not invalidate that it is an algorithm simulating an oscillator circuit.

All VA synths work with PCM format and feed the final digital sound into a DAC to render as analog. If you prefer to call it smoked based because it uses stored data in PCM format, so be it. The only subtractive synthesizer you will find that does not fit your broad definition of being sample-based will be true analog synthesizers.

You are correct that some sample-based keyboards have VA synth features, some fully complete rudimentary VA synths. The Roland Juno Stage has sine and square waves, LFO, resonance filters, ring modulator, and low, band, and high pass filters. There were users who used those features to create synthesizers leads with no connection to sampled sounds in the keyboard. This is a basic VA synthesizer embedded in the keyboard.

A pre-cooked noise oscillator is not VA. The loop point can become audible. Also, the pitch change artifacts will be for sure audible in upper and lower registers due to a lack of samples and parameters.

My humble guess is AL1 in Kronos is quite the reverse of what you just posted.

You can persuade yourself that Roland's SNS is VA but from my point of view, it is a joke.

It is important not to conflate the behavior of looped samples of an acoustic instrument whose sound decays over time like a piano with the behavior of a periodic function like a sine wave. If the wavelength of a sine function is L then sin(x + L) = sin(x) for all x. If you store one period worth of data, each loop is just another period of the sine function, and the values that need to be generated in each looped period would be the same discrete values as needed for any other period. Storing them for a single period so as not to have to recompute them will not give rise to detectable looping.

Specifically, if you have the value of sin(x) for some x there is no benefit to recomputing sin(x+L) because it is the same value. Just using the value for sin(x) will not contribute to noticeable looping effects. In fact, I would argue that computing all of the values in real time would be more susceptible to detectable looping if variability in computing time for the floating point computation of the sine function contributed to any timing jitter.

There just is no justification for considering a difference in detectable looping between generating the same value repeatedly and generating it once and storing it to be reused instead of regenerated.

I should add that the above is a simplified description because a particular wavelength may not be divisible by sample interval. This would require also using periods that are integer multiples of wavelength but this can be engineered. It is why samples of organs can be looped without detectable looping in the final sound.

But we don’t actually know how Roland implemented the virtual oscillators in any case. The quoted example above trying to analyze it by inverting the left channel and summing with the right channel canceling the sound is nothing more than a test of whether the sound is stereo or mono. If the phase inverted left channel cancels the right channel it just means left and right have the same mono signal.

You didn't read that post carefully then.

The main argument is about supersaw and noise. I think you know how pseudo-random number generator works and I hope that you know what supersaw is.

The test was taken twice and there are two unique recordings for noise and supersaw. If the audio is produced algorithmically (supersaw and noise) then there will never be a complete cancellation as the two signals will always vary.

Yeah, I missed that they were coming from different oscillators. If pseudo-random sequences used for both used the same seed, they still would be the same sequences, but I consider this technical implementation discussion to be less relevant. What matters is the sound quality of the sound engine and its programmability. It is a virtual implementation of an analog synthesizer based on the controls.

I have used a Roland Jupiter-50, Akai Miniak, and Nord Stage. Sound quality of the Jupiter-50 supernatural synth sounds are excellent, at least on par with the VA synth sounds of the Nord and superior to the Miniak (which has a somewhat more metallic, less warm sound in my assessment). I would not want to design sounds from scratch with the interface of the Jupiter-50, but it is fine for modifying existing sounds. Built-in synth sounds in the Integra-7 are more raw than in a Jupiter-50– basically it has the tones that in a Jupiter-50 are layered to make finished sounds built in to the J-50, but the Integra-7 lacks the finished sounds. I thus would be leery of getting an Integra-7 for its VA synthesizer, but for a different reason.

If what you want is Supersaw pads, why not just get the real thing, a used JP-8080, the rack mount version of the VA synthesizer that introduced the world to supersaws? You also will get physical knobs to twist instead of menus to traverse. A used JP-8080 should be much cheaper than an Integra-7 as well.