need help with rhino

Official support for: bigtickaudio.com
Post Reply New Topic
RELATED
PRODUCTS

Post

hi all

can someone walk me thru the process of importing dx7 sysex into rhino?

i am trying to import my existing hardware dx7 patches, and they are not coming across correctly. here is what ive done -- i dumped a bank from the dx7 into sonar's sysex read function. i experimented with two sonar options, "generic dump" and "tx7 32 voice bank dump" (there is no dx7 dump option). in both cases, the resulting dump is 4104 bytes; i think that aspect worked correctly.

rhino reads both dumps, but the sounds are garbled. the patch names came across fine, and the patch info appears to be getting thru. i cannot pick up exactly what is the root cause; it sounds like it is related to envelop timings and relative osc volumes.

any thoughts?

Post

well...it sounds like you are doing it correctly. The file size is the same as a standard DX7 sysex file. You say the names are getting thru, so it is importing into Rhino.

I have found it depends on what type of sounds you are importing. If there are long envelope settings on the orginal dx7 voice, it will take some tweaking once you get it into Rhino. Also, no mod settings ( vibrato, aftertouch, ect) from the original DX7 will be read.If you try importing ( as an experiemnt) quick envelope type patches, and see what it sounds like, this may give you ( and us) a better indication.

TimC

Post

an update on my experiments --

i also downloaded and attempted to import a bank from a popular dx7 web site, with similar unexpected results. the web bank has fulltines -- and we all know exactly how that should sound.

in both cases (my sysex and the web download), only 4 of the 6 osc were being read into rhino's mod matrix. there was not an obvious pattern; the two missing seemed to be "dx7 algorhythm" dependant. all six osc's were in use for all imported patches.

would it help if i post the original sysex and the rhino fbx ?

Post

pummel wrote:an update on my experiments --

i also downloaded and attempted to import a bank from a popular dx7 web site, with similar unexpected results. the web bank has fulltines -- and we all know exactly how that should sound.

in both cases (my sysex and the web download), only 4 of the 6 osc were being read into rhino's mod matrix. there was not an obvious pattern; the two missing seemed to be "dx7 algorhythm" dependant. all six osc's were in use for all imported patches.

would it help if i post the original sysex and the rhino fbx ?
As a start, can you even import and use the original DX7 ROMS ? They are available at a few different sites for free, just do a google for 'DX7 banks'

I was able to get all 4 roms to load with no problems. Not being familiar with the original DX7, I
don't know what might be missing on import to Rhino.
But I do know that they sound good with no extra noise
or glitches. Also keep in mind that both Big Tick and Tim have stated that the DX7 import to Rhino is still in beta,it's not a finished feature.

Cheers 8)

Post

thanks for the help, guys !!
Not being familiar with the original DX7, I
don't know what might be missing
dougby, are you sure they loaded correctly for you?

rhino did in fact load both test banks for me, and they sound "ok" as is; they just do not sound correct. i am not looking for an exact match, but i was expecting them to be recognizable.

i think the issue is that each of my imported presets comes in with only 4 of the 6 osc's. i know that is not right; it is extremely rare to find a dx7 preset that did not use all 6 osc's.


would you be so kind as to check one or two of your dx7 rom import presets to make sure all 6 columns of the mod matrix have entries?

Post

Pummel,

This is very likely to be a bug in the dx7 import feature (which is indeed beta - I wouldn't have released it, had it not been for the insistence of some hardcore users out there).

It'll definitely help if you send me the offending sysex files.

Cheers,
'Tick

Post

i have imported many, many sysex files into rhino from Tim's website and also the dx7 uk site.
the only problem i've had is the well-known click on release with envelopes that were already at 0 levels at the release point. and that can be fixed by tweaking (usually all 6) amplitude envelopes to slightly above 0 for some release volume.
i've imported many of the same files into fm heaven. with either import all 6 operators come across.
now they don't sound the same in fm heaven as they do in rhino. fm heaven as an emulation is more accurate in recreating the dx7 sound, but usually i prefer the richer, more complex sound of rhino on these banks.
very rarely (maybe 1 preset in 5 or 6 banks) there's a preset that doesn't function or in the one case of an fm bass patch where the creator tried to duplicate a patch from another Yamaha fm synth, and spiked my cpu to 100% before it could make a sound.
it sounds like something else is going on. i'd suggest downloading rhino again and re-installing it.

Post

yea, maybe it is my installation.

i had been testing the 1.08 demo rhino; when i purchased, i just installed the 1.0x->1.09 patch rather than downloading the complete 1.09 version.

i will flush and refill; hopefully that will fix it.

Post

ok,

i reinstalled rhino, and my issue remains.

it took me a while to get my head around the mod matrix; i understand it now, and the algorhythms are loading correctly with all osc's being used -- my bad.

however, i dug down deep and ive got some ideas as to the root cause of the differences -- i am just wondering why i am the only one having difficulty.

ive got two files to share now; the first is a sysex dump from my dx7. the second is a rhino bank with two versions of a patch, selected from the sysex dump: (FAST ARP, as imported) and (FAST ARP, tweaked). the second version sounds almost identical to my dx7; close enough for illustration purposes. the import sounds muddy and mundane; the tweaked version has growl and bite, similar to the live dx7.

for whatever reason, it seems as though *my* rhino installation is not importing the envelops and mod matrix levels to "reasonable approximations". is there a dx7import.ini file or something that i need to update?

also, is there a place i can post the two files?

thanks in advance for all the help guys.

[/code]

Post

pummel wrote:ok,

i reinstalled rhino, and my issue remains.
What are your computer specs? Maybe the problem lies there?

I'm using Rhino on:

Athlon 1600+ w. 1Gb ddr ram, dma133 IDE hd, ATI Radeon 128Mb card, 2x 19" monitors using Chainer and Cubase SX 1.06.

Post

nah, i do not think it is a computer hardware issue.

its related to the interpretation of dx7 sysex, and its subsequent translation to rhino-speak; and i cannot find the resource file to tweak the settings.

i am planning to take this issue offline, with tic himself. perhaps there is something i can do to help everyone.

Post

pummel wrote:nah, i do not think it is a computer hardware issue.

its related to the interpretation of dx7 sysex, and its subsequent translation to rhino-speak; and i cannot find the resource file to tweak the settings.
If that was the case though, why are you the only one having problems? That doesn't make sense.

Good luck :?

Post

i sthere a setting for the mod matrix where the values are linear or exponential? if i remember the dx7 was fixed at exponential, but it's fuzzy, but i thought maybe fm heaven had a choice of linear or exponential values -- or was that rhino?

wrong setting interpolates values way wrong.

envelopes were pretty close I though except for long drawn out closing out too soon in the translation.

Tim Conrardy helped Tick with the importer. He might have a clue. Otherwise just keep telling yourself -- it's only a beta, it's only a beta.

Post

yes, rhino has the mod matrix lin/exp switch; ive already tried that.

thanks guys for your continued interest in this issue.

please note, "having a problem" is a relative term; maybe i am too much a purist. I realize rhino's dx7 import is beta, and i think tic/conrardy (and others) have done a very good job so far; many would consider it good enough as is. however, i want to help tic take it that final step; from 75% to 95% accurate.

here is a little teaser just to show how serious i am about this...

FAST ARP1 ====================================================================
|------------------LFO---------------| mod osc
algorithm feedback | wave speed delay pmd amd sync | pitch sync
9 6 Tri 23 0 11 0 On 2 On
===============================================================================
S freq +- ----Rate--- ---Level--- kbd --Curve- Depth Kbd Out Sens
C F 1 2 3 4 1 2 3 4 brk L R L R Rat lvl V M
6 r 0.5 0 -6 75 49 24 44 90 99 94 0 E10 -lin -lin 0 17 0 99 0 0
5 r 0.5 0 7 70 27 16 40 90 99 95 0 A-1 -lin -lin 0 0 0 75 0 0
4 r 0.5 0 -6 70 27 16 39 90 99 94 0 A-1 -lin -lin 0 0 0 82 0 0
3 r 0.5 0 0 64 40 23 50 99 97 94 0 A-1 -lin -lin 0 0 0 99 0 0
2 r 0.5 0 7 59 32 26 36 94 99 94 0 A1 -lin -lin 0 5 0 86 1 0
1 r 0.5 0 6 65 99 99 51 99 99 99 0 A-1 -lin -lin 0 0 0 99 0 0
===============================================================================
---pitch EG--- Transpose CheckSum
Rate 99 99 99 99 A2 0x25
Level 50 50 50 50

[EDIT]
oops; well, we lost the formatting in the post, but the info is all there.

Post Reply

Return to “Big Tick”