Acxel II

[BobTheDog]

Well we are getting a bit off topic but:

The time it takes to transfer to the pacarana is very small, the stuff I have been working on is instant.

If you are transferring loads of samples or data into ram it can take a time as you would expect, this is only done once, if you change other things then the whole lot is not sent again only changes.

The longest I have had to wait with a complicated timeline is a couple of seconds.

Also DJ is not a Kyma user.

[xjscott]

The product brochure seems to explain what it does fairly well:

http://www.acxel2.com/pdf/Acxel2Brochure.pdf

I assume that the final product manual itself is not available because it's not yet finished.

Scanning through it, we have a custom DSP chip with a set of cells, each which can do certain signal processing functions. The cells can be interconnected to form patches. The 256 or 1024 cells are all that are available and are shared among voices. Voice management is done by an separate general purpose CPU. Output is to a quad arrangement.

This is very similar to the family of synthesizers I worked on back in the 1990s. We had 128 "audio processing units". Each unit could be output any of 128 submix busses, each which could then be the inputs to any number of other audio processing units. Outputs could also go to physical outputs, which were 16 channels, divided into 4 sets of quad outputs. All units had built in quad panners with a radius and direction setting, these could output not just physically but to the submix busses which could also operate as quad.

Using this system, you could build a 128 voice sampler, or reverbs, choruses, flanging, multistream real time 3d positional audio, physical modelling, PPG wave sequencing, additive synth using any waveforms, compressors, limiters, sample rate converters, wave players from sample RAM, sample ROM or streaming from host system memory, or any combination you could fit into the 128 voices. It was pretty cool for something available in 1996 and that cost $10-$20 to manufacture. Never sold any though.

[NoiseLab]

[TekniK]

[TekniK]

[NoiseLab]

I was talking about the Acxel 2!

[seraph]

[TekniK]

[GovernorSilver]

[pguilmette]

Hi guys
Thanks for your comments. I will try to answer to them in the following posts for the technical points. My colleague Robert will shortly add Post for the Support issues.

[pguilmette]

We will shortly add new ones on Acxel2 website. As Teknik said, the example on the website were no demos but example of resynthesised sounds with Real Time modifications at different levels, not very musical wise, I agree.
More to come very soon.

[pguilmette]

We understand that our very initial customers are very knowledgeable about the synthesis techniques.

The Acxel2 is designed as a very intuitive instrument.

We understand that any user has its own experience about synthesis modes, some others use presets and prefer to spend most of their time in composition (in fact they are in majority). The Acxelink mode is the native Acxel Synthesis mode, we build different macros that represent higher level (eg: Frequency Modulation, Additive Synthesis, etc.) and represent standard synthesis technique. All of this to make things simpler synthesis models, to be used in respect to users experiences: expert-sound-freak in one side, Keyboard composer, … and also alternate intrumentist (using another instrument to control the Acxel2).

We will add to the User forum, access to in deep Tutorials: Resynthesis, concepts, synthesis modes and how the Acxel2 can emulate them, etc.. We will make extensive efforts to help the Acxel2 users to use their instrument at its maximum of potientialities.

The access to the users forum will be unlimited in time, the 3 years term is for special download sounds - macros, and other priviledged support etc..

[pguilmette]

The Pentium into the Acxel2 Rack/Studio is simply a controller (for User Interface and high level control like envelopes, etc.). Then it is really different to a PC/MAC architecture as the ACXEL2's RPP processor does 99% of the job.
The Pentium option is simply useful for System Networks, Multiple Screen and Windows Supports and other High Level Software.
In fact the internal architecture handles a standard bus interfaces so that it perennity is garanteed.
RPP operating at around 200 GOPS (Billion Operations per second compared to 1 – 3 GOPS for a P4, or a DSP) replaces a high number of DSP. This means that equivalent performances would require between 50 and 100 processor put in parallel, this without considering the Overhead of this type of processing network.
The RPP processor handles the Synthesis and the Parallel Multiple Real Time Analysis/Resynthesis.
The Acxel2 is still cool technology, no fan inside.
Then each current RPP handle 250 cells (actually 16 voices – NOTE that this is a software limitation, in fact the hardware can 64 voices for 256 cells).
As TEKNIK said, ALL the process is manage in REAL TIME in the Acxel2, this includes: Configurations, Controls, Analysis, Resynthesis, etc..

[pguilmette]

This limitation is not a system limitation and is temporary. We prefer not to officially announce it in advance.
The definition of the Cells term is different from the Acxel-1 to Acxel2. In the Acxel1 it was Additive Synthesis oscillator element. In Acxel2 each cell is a multiple processing unit with multiple signals and controls inputs.
In fact the current Acxel2 architecture is highly scalable and supports 20 times more cells.
Any system working with 1280 cells (or 256, 512, 768, 1024) will be upgradable to at least 5120 before the end of this year with Modules that the user will be able to install himself (or if he/she prefers at store).

[pguilmette]

Because we received a lot of spammers who attemped to register in very large number into the Acxel2 forum, we blocked the registration until our webadministrator will filter the access to allow access to registered users (and discard repetitive IP# registration).

We will unlock the registration on wednesday 06/06. For the one who would like to register, please send an email to "webadmin@idarca-audio.com" first and proceed to your registration, we will confirm your registration.

Sorry for the inconvenience.

[Robert]

Hello dear friends musicians,

Just a few words to thank you for visiting our Acxel2 website.

I wanted to thank you for getting our attention to our OnLine Priority Service ... it doesn't seem quite clear.

As at Idarca-Audio we're all proud to be musicians and users of audio equipment, we are convinced that an exceptional technical service will be a large part of our success into musicians community, and we will work very hard to please all our musicians.

So we just want to confirm that ALL Acxel2 registered musicians or audio professionals will have access to a free technical service (Acxel2 technical standard service). Our standard technical service will work as fast as possible within a 24 to 48 hours delay, depending on the workflow.

As a musician, you understand that when on tour or into a studio recording session, technical problem could
occur and must be resolved as quickly as possible. It's precisely for this kind of urgent request that we created the OnLine Priority Support Service, which allows the subscriber to obtain a unlimited number of priority contacts to technician within 60 to 90 minutes, 24 hours / 7 days.

We had originally planned to offer this option only for our STUDIO model. However, after discussing with several professional musicians, we were convinced to allow everyone to have access to this optional service ...We may even sleep less, but surely gain many musicians confidence.

Don't worry, we love to be criticized, it's the only way to get closer to excellence!

Take care every one,


Robert Beriau
Sale & Marketing Vice-President
Idarca-Audio Inc

PS: watch the acxel2.com over the coming weeks for sounds and tutorial video to be added

[seraph]

Hi Robert
welcome to electro-music.com

[DrJustice]

[pguilmette]

Not exactly
Each group of 256 Cells handles 16 (64 in next software update this summer) voices, or 64 instruments or notes of polyphony.
Each Cell is a processing unit that can generate a single signal, can receive up to 5 simultaneous controls (amplitudes, frequency, phase, morph/waveform, filter) and up to 3 signal inputs.
About the signal inputs, the Acxel2 system can have from 4 to 16 signal inputs for each group of 256 cells (4 standard, others are optional). These inputs can be directed to the cells signal inputs.
Each Cell controls its own signal that can be combined to other cells to build an instrument (example: in resynthesis with additive and noise generators).
Basically the 256 cells system has up to 64 voices (a combination of notes and instrument). Example a 20 cells sound will have 12 notes of polyphony. NOTE: the voice allocation is dynamic then the resources applied to instrument can be redirected in Real Time where they are required.

[pguilmette]

If you use Acxel resources in wavetable/sample mode, each cell handles its own multiple samples (in fact an envelope is defined to control a multiple sample / waveforms sequence).
Also my example in previous post about 20 cells. Be aware also that each cell can replace 4 oscillators, then a 20 cells means 80 oscillators (or a 80 sinewave/noise/other preset waveform additive synthesis sound). Then a 256 cells system is equivalent to 256 simultaneous sampling units or 1024 oscillators or a combination of them.
A resynthesized violin, for example, requires about 30 sinewaves (or 8 cells), then you have 32 notes polyphony for this sound.

[TekniK]

[pguilmette]

Nice guitar group Teknik
BTW your 6 friends can simultaneously connect into Acxel2 Inputs and control their sound (as instrument or effect)

We just unlocked the registration into the Acxel2 forum.
Until the Spam issue is resolve, we will unlock registration each Wednesday.

Have a good day

[DrJustice]

[DrJustice]

[xjscott]

Great information thanks, and the service thing sounds reasonable for what it entails.

I understand that a cell can be in a mode to generate 4 waveforms. Does each of the 4 suboscillators have its own fine pitch, phase and amplitude control, and if so are they independently controllable with envelopes? Or is there only one pitch and amplitude envelope control possible for all 4 oscillators together which then stay in fixed relation to each other once configured. I'm kind of asking if the 4 oscillators are prerendered into a table buffer, or if they are generated live. The reason I am asking is that a complete real time phase vocoder/resynthesizer implementation would of course normally require independent control of all 4 subcomponents to count these out as 4 independent timbre partials, otherwise it would be more like 256 cells really only hold 256 partials rather than the potential 1024 suboscillators that have been referenced in the previous hypothetical resynthesis polyphony count.