Posts by tonstudio96

    Right, but as said, I know the limits of wavetables since we use them in SDR, RADAR and LIDAR for different purposes. Regarding sound generation, these limits create a lot of artefacts, which in some cases are the reason for the "good" sound. Depends on the way you access the table and (pre)process a wave.


    And yes, I know the microwave - I even tried to emulate it using also wave table synthesis:
    http://96khz.org/doc/vasynthesismicrowavext.htm


    At the moment, I am evaluating a new way of WTS, which is first possible with the current technology and available at a reasonable price. It requires some processing power and tricky resource management, but in some weeks I shall be able to present some results. Currently I am waiting for the new PCB. Since I plan a release also for B2C, it might be possible that there will be someting available in summer. Maybe as a companion to the Virus.

    Right, It can be made, but only if the hardware is similar (enough) and this is not always the case. To reproduce a sound, it might be helpful if users post hardware settings of their desired sounds describing the routing, starting from the OSCs, settings for ADSR, Filtering and their behaviour in the time domain. Often it is possible to gather that information from the knob settings of real analog synths and / or MIDI dumps if available. The biggest issue is routing: Analog synths with free wiring capabilities offer an unlimited umber of possible routing variants which is even hard to emulate in my hardware - an I am nearly able to program anything.

    Hm, I have a Virus B too and also upgraded to the 4.9 using the 1024 patches. For several reasons, I never changed patchs a at my virus sind then, so I cannot answer it precisely. These 1024 patches were introduced by replacing the built in (fabulous!) demo song, once available in the Virus B makeing use of the flash ROM. AFAIK this does not affect the normal behaviour of storing settings in the user banks. Please correct my if I am wrong.


    But AFAIR, there was is detailled desciption bundled with this OS update and the Patch Bank.

    What do you want to do in detail? Typically you will use one of secondary outs of the Roland to route particular instrument channels to the virus' audio input and configure it appropriately. You might use e.g. a vocoder - just as an example. AFAIK the 1010 has some choir pads. (I do not have one anymore, I use these traditional Roland sounds from within my VA76).

    Appears to be a MIDI overrun issue, because MIDI USB Devices might send too quick. Also WIN USB issues can take place. I have some interfaces here, driven by the LAPTOP and hardly can achieve the bandwidth, possible with the particular device. When I drive my gear (also Virus) indireclty via my fast midi swich there is no problem, neither with data nor SYSEX.

    Should be no problem, but with fixed WT - where is the aspect of individual sounds?


    Another, and possibly better choice, was a multichannel IFFT with coefficient control. One could Analyse any waveform and use it's LaPlace-Representation. Reconstruction in real time is easy with DSPs when using trigonometric synthesis like cordic or exponential functions (you get an octave shift of a sine wave my multiplication with itself).


    I already did that with an old 56301 DSP.


    With current DSPs, a pretty large number of channels = frequencies is possible without creating too much load for the ALU.

    That is the Point: As Long as you have a generic function, you can calculate any Point for any Phase in your Destination Domain, without any error (taking a good reconstruction filter into account when becoming analog later) so a simple Resolution with one value per sample frequency is enough.


    But any wave stored in a ram requires a at least a multiple of it to be replayed perfectly and a nearly perfect pre filtered Trigger wave (in fact the frequency of the note). For straight Forward frequencies, like a 12 tone System, one might prepare the tables appropriately for this but with synths and things like vibrato, it is not possible.


    The only way in a pure digital system is DDS and ignoring Nyquist by accepting several issues.


    For a 1024 dot wave for example, and a 1% granularity of the vibrato, already 100 Mio clock cycles are required for a 1kH tune. This is possible in Hardware, but with PC-Synthesizer and it's limited number of samples / sec this might get pretty dirty.


    One could think of a hybrid synth with an analog section for the wave table voice, acting on e.g. 1024 times the required frequency of the tone and driving the DDS-wave table. That would be smooth. I am using such things in my old PLD organ, which creates analog sine like sounds from a digital Domain, but resolution of the System was too low at that Point of time. I am only using a 64 dot "wave table" with fixed waves.

    What resolution would you suggest for wave tables? The sample frequency? The point is that digital synthesis operates either with DDS or partitioned (indirekt) frequency Synthesis. I wonder if this is realized in a DSP based synth and I wonder secondly if it makes sense. But only of Version two you might be able to produce an artifact free wave when looping.


    I am using this in my FPGA Workstation by applying a common DDS for the frequency generation and a subsequent PLL design offering 96.000 times the frequency of the tune.
    http://www.96khz.org/htm/wavetablemodule.htm


    With common techniques in DSPs this is not possible because it required extraordinary sample rates or you have to live with the artifacts caused by the unintended Phase shifts. For Details of this issue see here:
    http://www.96khz.org/oldpages/limitsofdds.htm


    So the vendors started to use generic functions to shun this issue when defining waves with harmonics of high bandwidths. So I am not sure if that was an increase of quality.


    Do you have any Information, how Spire / Serum do that?

    Impressive! A 3 year time out in Music making. This even beats my time out :-)


    BTW you should consider using a more detailled thread title than just "read this" :-)

    [quote=Honestly, second hand market probably hurts Access the most...[/quote]


    Do you really think so? IMHO People who do not want to spend Money on a current synth moreoften go with a japanese device.


    I think the market for Hardware synths is more and more saturated. SW Synthesis on PCs is the real Problem.


    HW manufacturers generally have increasing problems and search for new and cheap solutions the same time.

    I guess the primary sound could be done easily using similar procedures as with pipes, possibly some quicky freqency modulation at the very beginning is tricky to emultate the transient frequency behaviour.


    What is hard to create with all these "massive" sounds are the resonances and reflections of the environment. With such a strong and long durant sound, the whole ship will start to emit sound shaping the spectrum significantly. For the interpretation of the sound the hearing distance becomes very relevant, similar to a soprano voice, where you do not hear only the voice from the mouth, but resonances from the head and abdomen.


    I tried such effects with strings of harps and pianos but this requires a hardware where voices have an impact on adjacent (at first passive!) voices and this usually cannot be done with common synthesizer architectures. In fact if you hear a piano, you do not hear the string, but it's impact on the wood and the other strings, so this is pretty complex.

    You want to emulate the C64 sounds I think (and not use the Original C64 or bend it's circuit), right?


    Well there are some emulations out in VHDL and Verilog language to be placed into programmable hardware like PLD and FPGA which replace the original SID chip of the C64 and have a pretty good documentation. From that you can learn how to control it and possibly emulate it's sound capabilities virutally using the virus hardware.


    Basically the SID has the classical sine, triangle, square and a noise generator and some rudimentary ADSR-behaviour, cut off and such - so this can be done surely.


    8 Bit-reduction won't do here much since the limitited bit accuracy is a minor issue. I am not 100% sure but maybe it is possible to recreate this effect with a very low patch volume or at least processing out channel only and reduce the volume appropriately (not sure about that).


    The typical video game sound of that time especially with the Commodore VC20 and VC 64 was marked by some tricky usage of the resouces to e.g. get more voices out of the sound chip. A very common thing was to quickly roll over / swap the notes of a chord in order to "waste" only one physical voice. This can be done outside with MIDI or inside the synth with an arpeggiator function. Also the noise gen was used and the cut off frequency was adjusted that way that it meets the current pitch.


    Another (even more tricky) thing was overloading the internal ADC by to high amplitude causing a "warm" clipping - as far it is true, that clipping causes warmth :-)


    But what was with a total weird idea and buy a C64? There are MIDI-interfaces available