Kyma Forum
  Confabulation
  Yamaha DX7

Post New Topic  Post A Reply
profile | register | preferences | faq | search

next newest topic | next oldest topic
Author Topic:   Yamaha DX7
carlos alberto augusto
Member
posted 17 June 2001 11:56         Edit/Delete Message   Reply w/Quote
Has anyone developed a Yamaha DX7 Sound for Kyma?

Cheers
Carlos

IP: Logged

photonal
Member
posted 18 June 2001 11:25         Edit/Delete Message   Reply w/Quote
Hi Carlos
There are some FM examples in the "wavetable synthesis" section of the examples folder.

IP: Logged

pete
Member
posted 18 June 2001 12:54         Edit/Delete Message   Reply w/Quote
Hi carlos
Although Kyma has plenty of power , to try and do a direct emulation of the DX7 would be inefficiant. The DX7 has 6 operators and kyma has oscilators that are FM ready and could be configered to emulate any one of the DX7 algorythyms. The act of switching from one algorythym to another in one single kyma sound, would require 14 oscilators, and at any one time, 8 of them would have thier outputs switched off and would just sit there eating up prosessing power. This may seem like a minor waste but when each oscilator needs a multi tap envelope and every note of poly that you add increases the waste, it becomes too large. What would be a better solution would be if each DX7 algorythym had its own kyma sound.
I haven't made one but if you want more details on how to do it your self, let me know and I'll try to explain it. Remember that you will not get the buzzes and horible noises that the DX7 sounds had, as the DX7 was only 8 bit (kyma is 24 bit) and the sines had only 256 samples (kyma's standerd sines have 4096 samples) I believe.

IP: Logged

carlos alberto augusto
Member
posted 19 June 2001 10:09         Edit/Delete Message   Reply w/Quote
Hi all,
Thanks for your messages.
Thanks in particular to Pete. I see what you mean. I've made various unsuccessful attempts to build a sound to reproduce a single DX7 operator (you might call it a KX7 or a DX7 XXIV!) and yes! I'd love to hear your comments and suggestions.
One of the reasons I'm interested in this is exactly because I'm curious on how a 24 bit version would sound vis-a-vis the 8 bit cousin.
Of course, it's always possible to use Kyma to dirty things up a little bit as well, if needed...
Cheers
Carlos

IP: Logged

pete
Member
posted 19 June 2001 14:04         Edit/Delete Message   Reply w/Quote
Just a bit at a time.
If you look at the algorythym 1 where all six operators are in a straight line . 1 feeds 2 ,2 feed 3, 3 feeds 4 etc.
I think thats alg 1 but i dont have a DX7 here
First if you start with an empty sound and drag in the module called oscilator. Then open up the oscilator to see the paramitors. Then tick the frequency box under the modulation heeding. Dobble click on the space where the modules appier to redraw the diagram, and you sould see a new "constant" module appier. drag another oscilator module on top of that constant to replace it. Open that oscilator and do the same as before. If you keep repeating this you will eventualy see six oscilator modules in a row which looks just like the DX7 algorythym. If you rename these operator1,operator2 etc respectivly.If you go into the frequency field of operator 1 and type

(!KeyNumber+!Detune1) nn

Do the same to The other operators but using !Detune2,3,4,5,6 respectively.
Note that the detunes will need there range increased to +/- 60 in the VCS.
The DX7 had ADSDSRs instead of ADSRs so if you copy a "multisegment envelope" module and paste it into the envelope field of operator1 and it should look like a yellow square box like this.

[multisegment envelope] L

Add to this to make it read

[multisegment envelope] L *!Level1

Now dobble click on the space to see the new module and open that module.
In the field called durations type

{!attack1 s} {!FirstDecay1 s} {!SecDecay1 s} {0 s} {!Release1 s}

Then in the field called brack points type

0 1 !FirstBP1 !SecBP1 !SecBP1 {0-!SecBP1}

Then change both the start loop and end loop to 4.

Now the same has to be done to the other operators but of cause with 2,3,4,5&6 in the hot peramittor titles.

This is just a start and it will need velocity added, LFO added, the feedback parth on one of the operators added and calibration to the DX7.

but its a start
let me know how you get on.
Pete Johnston.


IP: Logged

pete
Member
posted 20 June 2001 12:33         Edit/Delete Message   Reply w/Quote
I missed an important bit.
you must put a value something like 10 in the MaxMI fields in the oscilators or it won't FM at all. These numbers will be used when calibrating it to DX7.

IP: Logged

metalbox
Member
posted 20 June 2001 14:57         Edit/Delete Message   Reply w/Quote

DX7.zip

 
Here are the DX7 algos in .gif format.

Enjoy.

IP: Logged

pete
Member
posted 21 June 2001 12:42         Edit/Delete Message   Reply w/Quote
Thanks Metalbox

Just what was needed and I now see that the algorythym I described (a string of six operators) dosn't even exist. It shows just how good my memory is.
But if we want to make algorythym 1 we need to use a string of 4 into a mixer and a string of 2 into the same mixer.
To make the feedback path around operator six we have to drag a memory writer module on the line between oscilator 6 and 5 .Open the memory writer module and type FB in the RecordingName field, set the CaptureDuration to read

1 samp

,tick the Cyclic box and Type 1 in the Trigger field.
Then turn on the modulation in operator 6 and replace the module feeding oscilator 6 with a sample module. Open the sample module and type the FB in the Sample field , tick the "from memory writer" box and "SetLoop" box ,type 1 in the gate field type default in the frequency field and type !FBLevel in the scale field.

It should now be obvious how to make all the algorythms as long as you remember to use a mixer anywhere, where more that one operators feed another (as operators 4 and 5 both feed 3 in algorythym7).

Does any one know if every operator has its own envelope or if some of them share them.

Also does any one have details of the paramittors of the envelopes as I think I may have got that wrong as well.

regards Pete


IP: Logged

carlos alberto augusto
Member
posted 27 June 2001 14:25         Edit/Delete Message   Reply w/Quote
Hi Pete,
Kyma X7 is slowly on its way. Everything seems to work fine.
Cheers
Carlos

IP: Logged

pete
Member
posted 28 June 2001 14:51         Edit/Delete Message   Reply w/Quote
Hi Carlos
I've just noticed that all the algorythms have something in common. At no time does a lower numbered operator, ever feed into a higher numbered operator. This makes me go back on what I said a few postings back, where I said that it would be too prosessor intensive to put all the algorythms in one Kyma sound.

But because of this common pattern it might be possable by using mixers buried among a string of operators, (OP6 on the left and OP1 on the right) with many interconnections, to allow all the algorythyms to be emulated in one sound.
Because all the connetions would travel from left to right, it means that only six oscilators would be needed and no extra feedback paths. It would need a lot of mixers under the control of loads of arrays , but it might just work.

I need to think about it some more but it would be nice.

Its a pity that the feedback path is not more consistant with all the algorythyms as this will require almost twice as many mixers.

IP: Logged

pete
Member
posted 30 June 2001 22:08         Edit/Delete Message   Reply w/Quote

KeemEx7.kym

 
Yes it seems it is posable to put all the algorythyms in one sound and here it is attached. It is still missing all the extras and dosen't yet have the feedback in it but it's a start.

Pete

IP: Logged

carlos alberto augusto
Member
posted 06 July 2001 14:52         Edit/Delete Message   Reply w/Quote
Hi Pete,
It works great.
There are a few parameters (the extras you mentioned) I still don't know how to implement I seem to miss the original manual. But this seems like an overall much better and workable start.
Thanks.
Cheers
Carlos

IP: Logged

pete
Member
posted 09 July 2001 10:04         Edit/Delete Message   Reply w/Quote
Yes Carlos
For the extras we need to know what peramitters are controlable by velocity, what peramittors are controlable by Keyboard number and what ones are controled by LFO. I've never had a DX7 or owned a manuel so it seems we are at the mercy of some kind hearted kyma user to put us out of our misery, like Metelbox did with the algorythyms.

IP: Logged

capy66n320user
Member
posted 10 July 2001 12:01         Edit/Delete Message   Reply w/Quote
PDF owners manual(s) may be obtained at:

ftp://byrd.math.uga.edu/pub/html/dx7.html

IP: Logged

pete
Member
posted 15 July 2001 18:30         Edit/Delete Message   Reply w/Quote

2KeemEx7.kym

 
Thanks for the manuel
I now see that there is a hell of a lot of extra bits. Most are quite easy to implement but very laborious as there are so many. It will need a DX7 owner to do a lot of tests to get the calibration and technecs. Of couse this project is probaly less atractive to a DX7 owner as the already have a DX7. Anyway here's a new one with velocity added and frequency paramittors more like the manuel.I've added sustain aswell and adjusted the levels to stop the cliping in the mixers.

IP: Logged

All times are CT (US)

next newest topic | next oldest topic

Administrative Options: Close Topic | Archive/Move | Delete Topic
Post New Topic  Post A Reply

Contact Us | Symbolic Sound Home

This forum is provided solely for the support and edification of the customers of Symbolic Sound Corporation.


Ultimate Bulletin Board 5.45c