WALDORF Ideas & things to remember Part 2 Ideas & Tricks

WALDORF… Ideas & things to remember

Ideas & Tricks…

how many voices can the microQ do and what about the DSP-power??.. go HERE!
UPAW wavetables.. how to make some..?
1) Use the UPAW assembler which is available as open source on the FTP-server.
2) Use the Polyframe (atari ST, was what is now sounddiver) Adaption georg müller created back in time for the Atari.
3) Use a Hex-Editor to code the SysEx directly.
but normally you can use the ones,that are already in: FM, and some more using sounddiver. upaw is a mathematical forumla, that discribes the waves and wavetable transition.. so it does not need space for waves.

since the next OS in the XT will contain the reserved wavetables to be filled up I assume most of them need to be upaws.. maybe some more are introduced.. maybe full cycle, because it is possible.. atm all waves are mirrored. even if its in the manual. there are no full cycle waves so far (current os is 2.33)

need more LFOs? set an Envelope to Loop between Attack and Decay1 and route it to the desired destination and set their rates to very low amounts (try the lowest and go ONE step higher)..
if you want to mod the speed: mod the Attack and Decay.. you can also set Attack or Decay to 0 to get a Saw (up/down).. and note: if you have both: you are modding the SYMMETRY.. so mod both and you have a speed modable LFO that can be very fast..
same of course for those who want even slower LFOs but set the Attack and Decay to high rates..! this also works in the Andromeda or other Synth with Loopable ENV’s.. like the XT.. aaaand: you can build your own Waveforms.. ;-)

>but the MW1 waves differ from that on the MW2 that way that they have a whole cylce not a half mirrored one!.. but this is not a problem I guess..
Both have half-cycle waves. The MWII can play back full-cycle waves, but all waves that we took from the MW1 are still half-cycle. I still don´t know why this limitation…? it is prepared but not yet able to do so…!!

wanna turn on all the lights (LEDs) of your Q?
press osc edit + mix edit when startup. welcome to the system testmode.
now turn the dial right to menu point 4…
..press play/ok and play with the dial to turn on/off every single LED.

self oscillation?
XT & Q selfoscillating filters
mQ needs some input to start self oscillation
(Rackattack) & Attack needs some input to start self oscillation

Analog filters need some input as well, but there is allways noise ;-)
But there it is a matter of design and calibration. Some PPGs do, some don´t, the latter is correct with regards to the service manual.

>Since the MWXT sends all it’s controls out via MIDI, can it be used to remote edit an MWI.
No, the sysex format of the machines is completely different. The cause: the sysex messages for realtime parameter changes are much shorter on the MWII/XT to keep MIDI traffic as low as possible.

I’d like a nicer way of the naming procedure.. maybe use the keyboard or like on the TG77 the phone-style of naming.. much faster than those things currently in most synths..
„drop out“-like CLICKING&crackling in the Q??: the problem is going away for most people when they’ve switched to „DAC Format 2“.
– adjustable DAC Format supported
try different settings (1-4) if any popping/clicking appears:
1.) Power off the Q
2.) Press ‚Global‘ and hold it
3.) Power on the Q – [** Service Mode **] should appear few seconds
4.) Release Global & Press it again
5.) Move to the last page in global-menu
6.) Select Dac Format
7.) Power off the Q by ‚Power‘ button

Also…
The Q now allows for separately controlling the sending and receiving of
Controller and Sysex data.
From the global menu, there is the control to have the Q send/recv
nothing, or send Controller or Controller + Sysex. Recv is „On“ or „Off“.
Check to be sure that Recv is „ON“. It MIGHT default to „OFF“.

IF you want to upload a new OS into the Q it might work only with this: power it up with peek+multimode buttons pressed.. it says: dump system now and shows a „thermometer/progress bar“.. also: slow down tempo of the sequencer, enable sysex in the sequencer.. there have been probs with cakewalk.. cubase seems to work fine (get a demo if anything fails..).. I had no probs with Logic,too..YOU SHOULD HAVE AN OS AFTER 2.15!! it is the OS thats works for all.. or NEWER!!.. and watch out you got the latest BOOT OS!
How about selectable envelopes for the Amplifier. Presently the Amplifier Envelope is fixed to the Amplifier but it would be nice to switch off the Amp env and use LFO or Wave Env..etc to Modulate the AMplifier. I think the present work around is to set the amp env to 0,0,0,0 and modulate it’s sustain by the above sources…plus the needed bipolar offset….however this is not obvious to most.
Also individual time and amp levels of the wave envelope as modulation destinations….maybe just the first 5 would be adequate. This will make ramp generator emulations and such more interesting.

WAVEs
Other interpolation strategies between waves in a table than linear, also as an option for a wavetable (let say : linear,polynomic,cubic splines,some weird waldorf suggestion etc.) .

just to re-tell it: I still would like full cycle waves!!..
and the rest of the ROM filled up with some nice stuff..
(then full cycle ones..)
-the noise source would be SO much more useful if you could shape its frequency response, even just a little bit. All you’d need to do would be to provide the option of putting the 6 dB filter after it instead of after the main filter. A BPF option on the 6 dB filter, if you could reposition it, would be really helpful too.. should be a source for FM, too. what about blue,red,pink noise..?

RANDOM NOISE,CHAOS in the OSCs (not only the noise knob!)Mark, me and others had some ideas of noise and chaos.. (the Morpheus has some nice Ideas in the LFO section!!) and to have it possible on a normal OSC.. cool.. but you know it is based on wavetables.. (I guess you know what this means..)
maybe it is possible to do random wave, anyway but I guess its hard work..
(if it is possible to find a workaround the Wavetable/UPAW…)

I suggested those chaos things like it is in the EMU morheus for LFOs, too..
the „random“ thing is improvable..and of course a smooth Random mode.. as seen on the Nord Lead…

in some cases a noise with a filter would do that job…

is there any change to get different types of noise oscilators for the Q and also for the XT in any future system update?
i would like to have all 3 type of noise like brown, white and pink.on the Q (maybe on the XT, too) every osc should have the option,to switch into different types of noise. so you could filter the different type of noise with a different type of filter…
I would like to see the Random LFO waveform more random (remove the
sinusoidal component at least :) and if the S/H LFO waveform was
modeled more like the one on the SH-101 then the MWII would sound much
better. When I was A/Bing the MicrowaveII against the SH-101 the better
S/H waveform was surprisingly apparent…I found it strange because in
theory in is not a difficult function to create nor a complex waveform?
Maybe the sample source of the MWII’s S/H waveformis the problem?

the OSC / mod matrix:

If „Fine Tune“ or „Unison Detune“ were modulation matrix destinations, then you could get really subtle in this area.

FM / Sync should be exchangeable (Master <–> Slave) eg.: OSC1 <–FM–> OSC2
MIDI vs. internal resolution
I think they both have their strengths. I like the non-quantized knob movements of a true analogue but I like the reliablity, convenience, and power of Analogue synths and some digital synths with MIDI…the Pulse MWI and MWII being three examples. I think the problem is simply analogue emulators are starting to get very good (and will only continue to get better and better) yet the midi 7bit 128 step per knob is sonically limiting.

Does Waldorf plan to use two double byte messages on the Q to get better resolution of the very important knobs:

• Cutoff
• Res
• Filter FM amount
• LFO rate
  what about to add more „space“ to the filters?

what do I mean?: well I think that 128 sometimes isn enough.. and that may be true for some other parameters as well..
(reso, etc..) what about have 1 or 2 or more high resolution windows where you can determine far better between 113 and 114
got it?
I’d like some modular resolution enlargement for better adjustment.. that has always been my problem.. (see Xpander!!! only 63 steps).. Res is always cool between 62 and 63 !!! hmm..)
maybe this is a possible workaround and I would not need it always.. so why not flexible and modable freely.. so there is not mod problem or so..

If not then maybe this should be looked at in the sysex implementation being developed.
..everything that has to do with FM adjustments and Cutoff/Res should be as hi res as can be done: but:
1. I must be compatible to the old 128 ..
2. if it4s be done there should not be too many sysex data.. and you
should be able to decide because it will increase the traffic…my XT is directly connected to one port but it shares the cable with another machine.. if it got too high…(traffic!!!)
3.there is a solution for the quantising: smoothing it…
but for me the more important (if you have had an Xpander you know what I mean!!) is that it needs to be 16 bit.(steps!!)

so this has do be done very careful.. if I can´t trun the knobs cause of the traffic…

LFO:

as the LFOs in the XT are fast (slow compared to the Q.. but 100 Hz is ok..?) perhaps an additional (or more) „slow“ LFOs (to reduce the need of DSP-power).. I was inspired of the little GECKO as LFO discussion.. maybe this is a good point for sequencer programmers (Emagic, Steinberg, Opcode …) for adding some sort of LFO that can be triggered by bpm and notes (1/4 etc..) sending out CC.. so you have an additional (slow) LFO! got the idea? maybe the XT itself could do another slow LFO.. also the Q may be able to do another 3 half speed LFOs?? (thats an internal thing then..)

btw: there is still some room to make them a little faster says Wolfram Franke (developer at Waldorf)!!

LFO speed – more ideas: tapping in speed maybe tap tempo function for arp-speed as well?

wishlist FX (XT):

modulatable FX (mod matrix..)… most wanted: dry/wet CC / play access params.
modulatable by midi clock.. (ecspec. delay) / BPM measure?? the Q already has it ;-) is it possible to include it as a mod. destination in the future?
Those things will show up on the Q. But on the MWII, it is not sure. The reason is absolutely simple: this destinations are monophonic while the mod sources might be polyphonic. And this structure isn’t that easy to implement.
However, realtime parameter changes for the effect parameters are possible (if not already implemented). They only don’t send out their values at the moment but Stefan said that he would have a look at this.

Yes, and a simple lowpass/highpass filter after the delay would be great, it would add signifigant flexibility to the delay algorithms…

VOCODER in the XT..! >
the RAM in XT, perhaps even in MW2 is enough. A vocoder does not need a lot of memory.

but. a vocoder is a bit :-) (too much) more to do for the DSP..

And a vocoder will of course need some calculations, if there is not enough left I will sacrifice some voices.
yes!!! this is possible.. I’d like to sacrifice some of those 30 voices for more synthesis stuff!!!..

But It will take some time because there are still other things on my list. (Stefan Stenzel)
maybe it could send/receive formant spectrum data as CCs

• If the Microwave’s DSP resourses could be focuced on a making a „hell of a“ single sound rather than distributed over 8 instruments then it may be possible.. Waldorf should expand on the MWII/XT & Q so that it has a non-multimode in which the filters can be stacked (means: 30×2 filters in the XT, 2×16 real Q filters)!!! thats what I call „hell“, in parallel/serial, 30×2+?? lfo’s, 30×2 osc …etc!! in the Q you can refeed the output via individual out back to the audio in.. and youve another complete filter & FX.. also possible on the XT.. but it would be nice to have that in the software.. much smarter.. and.. more stackable voices ;-))

It should be possible to feed an external signal in when in multimode and have each instrument on the same channel but with no Waves, Noise or Ring mod assigned in the Mixer. Just choose a cool filter for all 8 (16) instruments. set it to keytrack at 100% and offset each. Bandpass and FM will be cool. If you use FM, be sure to offset OCS B for each instument so that you can cover an octave or two. This will give pseudo Vocoding (interesting) results.

what about a TRACKING GENERATOR in the Q/XT.. as seen in my beloved Xpander?.. you can do nice things with it.. eg: scalings etc.. (would fit nicely with wavtables etc..).. also makes it much more „human“.. I used that one on lfo speed.. its really feeling sooo nice..

free assignable CC to each Knob of the XT & Q (..).. for use as aknob box / Controller for other gear..& local off for the CCs!! (so it leaves the internal sound unchanged).. ok.. if you use another free „empty“ multi.. thats also possible.. but would be nice..

more FILTERS
some more filter types etc.. like the Waveshaper and S/H Filter etc.. from the XT in the Q!!..?? and:

what about a comb in xt, too?
Sorry, impossible in the XT, not because of DSP Power but because of DSP RAM for the delaylines.!!
(Is this because the effects RAM is separate from the synth engine? I would gladly give up one of the three effects if the RAM could be put toward a comb filter.)

Yes it’s true It’s not always a solution to reduce voices!! for getting the comb thing it is an issue of RAM.. and this is only enough in the Q.. this is why I vote for another expansion „more RAM for XT + voices“ dunno if this is economicly ok.. but for me as a user and synthesis junkie it IS! but: It is NOT possible to add fast RAM to the DSP of the Microwave II/XT!.. so the sacrifice voices may be the only power saver..

The XT doesn.t have enough dsp ram for having the Q comb as filter, but maybe it has enough effect RAM to have the Q comb as effect ???
Sacrificing Multiparts / More Multis???
More parts don’t need more DSP power, only a little more RAM. On the Microwave, it would need much more RAM because of the wavetables but on the Q, it is fine. (This implies: Comb on XT is becoming harder to do for this 2nd reason!) BTW: it makes a lot of fun to use drumsounds from the Q and then you need lots of parts.

what about an odd/even harmonics filter?? plus adjustable harmonics (…not possible till now.. WISH!!)

What „neat feature“ would I like to see?
The S&H filter with a high pass or bandpass instead of a lowpass.

Ah- the K5k also has an Even/Odd Bias knob.Very handy!.. also.. what about other harmonics filter 3rd? 5th.. or combination??.. I loved that in the morpheus..more wishes..: and more like the waveshaper and some easy handy harmonics filter (the K2k / K2.5 has that,too and it is cool there, too) Ya- the K5k has a Formant Filter too, mainly to make up for some additive shortcomings..

Harmonics odd/even is POSSIBLE in the Q: You can do an odd harmonics filter on the Q by applying 100% keytrack to the comb- filter, matching the frequency and setting Resonance to 0, only odd harmonics will pass. Same settings, comb+ and only even harmonics will pass, which results in a tone one octave higher.
COMBfilter in XT
I talked with Wolfram about the comb filter in a XT (after being impressed by the Q). He told me that the comb is pushing the Q to its DSP limits and for that reason it would be ‚probably‘ impossible to port it to the smaller
DSP of the XT.The reason that comb filters will not work in XT is that they need some memory, and the MW2 does not have enough of this.

if it is possible to reduce the number of voices there could be a bit more power to do such things.. if it will really work? up to Stefan.. (& the DSP)

Stefan: As I said, no matter of cycles but a matter of memory, so sacrificing voices will not help.

About the voice sacrificing for power: well, W say no because they want it to be compatible to the 10voice version.. so the voice exp. will still be a voice expansion.. (anyway.. I still hope for using this to get some more power for synthsis.. but this is official!)

SACRICICING… for getting more..

it might be that even when it is reduced there is not enough power left.. ok.. but this is one of some ideas.. and sure they will implement some more filters / combinations (that make sense)..

another way: switch of the 2nd filter, too .. so we would have: less voices one filter uses power of Filter 1 and 2.

official answer here is: sorry, the 2nd filter is not modular, so it cannot be left out.!!!!!!!!

so again: you need to sacrifice voices.. imo it would be ok to go to 5 voices on the 10voice model.. switchable then!.. or simple: the 30voice expansion should be a 15 voice expansion plus synthesis maximizer.. ;-)

is it possible to get a room / reverb?

W answers: Perhaps in the XT, but certainly not in older MW2. The Q will surely get one.. one of the next updates..

what about the cool XT – BPF in the Q too.. you need another parameter (as the extra parameter in the XT) and you got 2 nice adjustable BPFs (modable quality/range then.. instead of using LPF & HPF..) and anyway: there are some filters in the XT that could be ported to the Q!!.. and I still want harmonics filters. ahh.. btw: I can do it with the comb!!

Mod’in filtertypes?? one thing that the sp808 can do that is even tedious to set up on my Nord Modular is that it’s stepsequencer can mod various filter types per step (ie choose a different filter type at each step). i don’t know if this is at all feasible on the Q, but I would really like it to be – it really adds life to sequences.i know that i can approximate this with routing/balance/LFO’s/env’s, but it would be really cool be able to choose filter type per step, at least for one of the filters. anyone else want this .. ?

some thoughts on the Q:
but I don’t really think that attack samples are possible..but what about algorithmic attack „samples“…
(as there is no place to store samples..?)

supersaw and the like.. hmm.. not on top of my list.. but just another idea..

I would like alternate things .. a bit far from standard oscs…
the formant filter thing sounds good to me.. what I love on the Z1 is the resonant OSC:
what is it?
a set of mod’able BPFs feeded by noise.. (as much as possible.. the Z1 has 5) well.. I guess you can imagine what you can do with this!! ;-).. hope this will be consider to be in the Q also..workaround: refeed a Q voice and use the now 4 BPFs plus the normal OSC set.. not really cool but..

then there are some nice ideas ranging from the clavias to grain (!!) things and wavetables (ok, it already has 2 complete Wavetables.. I’d like the XTs wavetables and editable ones also ;-) maybe some MORE FM/Waveshaping/… and mod’able.. means: waveshapier as filter and in the OSC section.. I just love to imagine that fun.. have it all in the OSC (!!!) section.. ;-) also: additive to be controlled by the knobs (some shift mode needed)
maybe look around those newer other competitors… there are some nice ideas to steal ;-)

and Wolfram said something about the Wave (2).. maybe there will be some sideproducts.. after that..??
..
I’d like to have some of these XT filters in the Q , too…(as said..).. as there are Wavetables and 2 Suboscs.. the Waveshaperfilter should be possible when using the ALT Waves.. also: S/H modable sampleratereducer plus filter as in the XT.. better than in the FX section!!

haha..there is an edit button to set some weird new params ;-) and thereis shift… ;-) …

MORe Q thoughts… 4op FM is cool.. I guess that could be done.. (DSP?!?!?!!)? also some slow LFOs maybe just one waveform??
I’d like to set the duty cycle of the other waveforms (for: sine,saw->tri.. =“Symmetry“)) via the PWM CC..

SOME MORE: XMorph: does a virtual „turn all knobs till they reach another sound“.. and: it already exists!! :-).. but some modules should be excludeable (eg: pitchmod etc..).. note: it does not morph the modmatrix and „switch“-type parameter do not morph.. they are switched at the „transition“.. currently it can be controlled by ModWheel & Pressure.. what about the modmatrix here??.. some parameter may be have some limiting / importance parameter for other morphing FX (maybe alternate curves).. and a chance to save a morph as a new sound

hmt (hermode tuning) which corrects all played notes to the real „correct“ clear tuning.. like: tuning the 3 OSCs as an accord.. disappeared from the MW2/XT(k) and Q/RaQ.. why?? (licencing issues & money.. I know..).. but it was cool..

another Q thoughts…well,
I’d like a more effective modmatrix.. it should be more „hot“
so I do not have to sacrifice 2 modslots for just macking it hotter…

and again the chaos/random things, the drumeditor for the XT and a fine tune mod (not sooo coarse)…

just a reminder..

and yes.. there nothing agains reverb.. even if its is „only“ korg type.. it is simply a nice sounding effect to mod reverb (eg mod amt and time) I love this in my Z1…

btw: and it has its own dedicated DSP only for FX so: why not?? maybe we could have some moadble resonator things.. (done with reverb)..
atm I have to use an xternal box.. and for THIS porpose I do not need a hi-priced mega Lexicon…

I would only vote against it if it blocks new synthesis function (as it does on the XT!!)…
but I always prefer onboard solutions for lazy ppl like me… and of course for live…

I do not agree with „only very good“ or „no reverb“!!.. you can call it cheapREV and everybody is happy.. there is always more than duality!!! (on/off!!!)..
some nice OSCmodes (valid for the Qs alternate Wave,too..) like superSAW or the like, spectral things, additive??

what about this in the Q & XT: I’d like to use the expansions power for modular synth / more synthesis power (additive? 4-8 OP FM, granluar?,formant things, harmonics…).. and having a „just bought / newbe OS“ including demomode etc.. to be exchanged by a real demomode-less OS to get more space for real synthesis things.. etc.. maybe a complete OS for turning the machine into say: additive synth etc… (c above)..

a local off for the knobs!

maybe you could choose „low Frq“.. LFO and get some DSP power for making
2 LFOs, too??

but always chooseable!!..

>This brings up another good point in terms of LFOs. If LFO speeds
>appeared in the modifyers (mdf1-4, in the matrix), then you could create
>an additional 4 LFOs. You’d simply use one of the 3 LFOs as the beginning
>speed and then apply any kinds of functions.

notes from Dr.G.Müller:
To get a full scale SAW or SQUARE you have to use

033 SawSync 1 startwave 00 for SAW
036 PulSync 1 startwave 00 for SQUARE

since the SAWTOOTH and SQUARE waves in the end of all tables use only 7
of 8 possible bits. I think this was some error in converting wavetable
data from the old PPGs or it was allready there.

By this you gain 6dB.

when dumping sounds: watch the (device)ID.. and there is also a broadcast ID.. other synth: most it defaults to 0.. so if in doubt use 0.. or the broadcast ID (not sure if a broadcast ID exists)

no more problems with IBK10 Control (needs all CC to be send for relative control!) is implemented now.

Deopfer Drehbank but it has no display and no Mac Editor (SHAME on you!!) and of course no relative Mode! but 64 Knobs! -does not need this feature.

a genreal Idea: All Parameters should be available through CC.. and if the 128 Midi Controllers are full use one as a Shift and you have got double, use as 2nd Shift and you get another 128´s Set… SysEx causes too much traffic.. mostly this is reality in the XT / Q..
I hope I get my 2ndFilter-Parameter as CC, too…
at the moment use play params or the mod matrix (any source in the matrix).. till it is implemented.works fine.. so it is not really urgent..

Yep… Wolfram has made it pretty clear that the Q is „yet another Waldorf synth“ which means that the knobs will (eventually) send CC’s that are compatible with like named parms on other Waldorf gear (ie, you can control a Pulse from a Q); the modulation sources/destinations are similar; the voice architecture is similar; etc…

PULSE:
I you want to burn your own EPROM you need this type:
256 kbit EPROM type 27C256 250nsec the OS is available on waldorfs ftp site!!
updates?? not possible!! the EPROM is full!! In fact, a couple of instructions could fit into the Pulse but the Pulse Plus
couldn’t be supported then. Then we would have two different OS’s for them which isn’t what we want.

> Is the 4-Pole the same filter as the Pulse Pluse in?
It depends, there are two versions of 4pole out, the later ones have the Pulse filter and the early ones don’t. Now don’t ask me how to distinguish them…

> I’m sure some of the patches created with the Randomize feature get stored incorrectly. I think someone mentioned this before. I tried getting round that problem by storing randomized patches straight away,
but it still happens. When I edit a random patch to make it more to my taste, what is stored is not always the same. In particular the envelope settings and maybe the wavetable / envelope seem to change slightly.

STEFAN:Yes, I think I found the bug, the wavetable was not loaded correctly. The fix will be in the next update.

>Would a true stereo input to the XT be possible in a future software upgrade, or is there a hardware obstruction to this feature being implemented?

It is planned to support selectable inputs for the external in of the mixer section, like stereo, left, right, … where stereo is only true stereo in dual mode and left+right in normal and unisono mode.

> I would support the call for alternative user tunings (if this is not already possible)… since I use quarter tones quite a bit.

Tuning Tables are of course possible in Mw2/XT, but we have not yet planned to implement these. Speaking about quarter tones, you can do this by setting the Osc keytrack to 50% .

> > Well, I also wish there were some „power-user’s“ front-end to editing wavetables, but most people do use some sort of computer to sequence or as a librarian, so I think I’d rather have more cool onboard features (like the effects processing, etc.) than have a hard-to-use wavetable editor built-in. No reason it has to be difficult to use. All you need is a screen like this:

Quite right, this is what we have in mind for editing wavetables in MW2/XT, but we will only have one wavetable edit buffer, and the screen will look like this:

0123456789012345678901234567890123456789
+—————————————-+
|Wavetable |Position |Wave coarse fine |
| 025 | 60 | 3XX 321 |
+—————————————-+
Plus the ability to copy, create or randomize wavetables.

…

>I vote for some basic UPAW editing (let say the FM stuff with some basic envelopes) for Onboard editing.

That’s a good idea. I had this idea for the big Wave in the past but the number of parameters just seemed to make it impossible. But having it limited to only a few parameters could make fun.

>> Randomize OK, but not cut & copy, it should be sufficient to store to a different location. ROM wavetables will also be editable, except the algorithmic ones. Soon…

The speech robot was introduced with V2.0 on the MW I, the MW II/XT has
it to, but only the built in 1-2-3-4-5/19-20 tables are working. When
loading data via MIDI in V2.14 the robot produces jsut garbage, sort of
invers babelfish. This has to be fixed in another firmware update.

The converter can be found in
/pub/waldorf/microwave/mwrobot/mwrobot.taz. In this TAR archive the file
mwrobot.txt contains some info on the synthesis method. The Sound chip
in the Atari ST has 3 voices plus noise, so this structure is emulated
by the MW when it receives a special SpeechTable Dump.

A robot wavetable has a magic long of $DEADBEEF (a upaw has $12DEC0DE)
followed by a wavenumber. That’s all. This wave and the following 7
waves
are used as parameter lists for the speechsynth. The format of this list
is:

60 entries, each 8 bytes long. The format of such an entry:

1.) Level Formant 1
0 – Max level, 8 – silence, sort of negative logarithmus to base 2

2.) Formant 1 Frequncy
this is in a linear scale

3.-6.) same for Formant 2 and 3

7.) Noise Level
this time the positive logarithmus, base about 1.414

8.) Noise Frequncy
„Pitch“ of the noise

The parameter list are generated by a hack which patches STSPEECH
and then sends the tables to the Microwave.

An example for such a list (part of it, at last)

sp12345:!speech wavetable „12345“
.byte 0,11,2,24,2,90,0,0
.byte 0,11,2,25,2,89,0,0
.byte 0,16,1,35,3,87,0,0
.byte 0,22,0,43,4,87,0,0

Perhaps Wolfram can compile this into the FAQ ??

BTW, you can reach me now directly as muellerg@diax.ch

Georg.

• Dr. Georg Mueller Georg.Mueller@physik.tu-darmstadt.de +

+ Diplom-Physiker http://www.physik.tu-darmstadt.de/~mueller +

There must be a way of implementing some form of wave-creation from the
front panel of the XT:
perhaps by just designing a start and end wave for a wavetable using some
basic additive synthesis.

Then there are all those lovely rows of knobs which could be used to
control harmonic ratios of a static wave…
I know that this isn’t a substitute for PC/Mac wave creation, but it would
be a nice touch and an efficient use of real-time controllers too!

> I understand Phil is working on a .wav import option for the uWavEdit which

That is possible, but because of the mirroring function, you must make
sure that you create symmetrical waveforms of the „.wav“ chunks. That way
the second half of the waveform will be ‚mirrored‘ and not sound all
choppy… That’s process is accomplished using FFT-iFFT (I believe,
correct me if I’m wrong).

I did some playing with this about four years ago on the MWI, with mild
success (my wife saying „I’m pregnant…“) but have since sold my MWI and
have a MWIIXT. I took a Pascal program sent to me by Georg Muller and
rewrote it in C and compiled it on a unix box and ftp’ed the file to my
mac to be dumped into the MW ;-0)

> however) Phil tells me the mirroring function which the uWave applies to

Then I find out that there is „speech synthesis“ built into
the MWI (and perhaps MW IIXT). For more info on Speech Synthesis, please
see ftp://ftp.waldorf-gmbh.de/pub/waldorf/microwave/mwrobot/ (or the
appropriate mirror at waldorf.ddv.com)

different Wavetable for OSC 2 and more:

at first if you want 2 independent Wavetables played by each OSCillator: make your own wavetable where one half represents one Wavetable etc.. so you have 30 Waves per Wavetable.. maybe this might be enough? but if you want more its up to waldorf…

just a thought that crossed my mind today in the morning while brushing my
teeth (is this unusual for musicians?): After hearing what could still be
done on the XT I was wondering if there would be enough storage for another
(queued) wavetable that the play pointer could point to after scanning one
wavetabele for linking several wavetables to achieve almost endless tables
with smooth transitions. However, I guess that is almost like asking for an
independent second wavetable for the second osc, which you already declined
for storage reasons…? The ‚linked wavetable feature‘ could be implemented
without introducing too many new parameters; the wavetables should be
stored to subsequent places (no arbitrary scans possible -> no need for
lookup tables) and the length in number of wavetables should be given. Then
the wave_env applies for the whole ‚long wavetable’… possible?

it can even do some things that the microwaves can’t

You might smell it but you won’t get it :-(
This topic was discussed 1 1/2 years ago. The reason why we can’t implement
that is easy: memory. This is also the reason why the multi mode is 8 part
multi timbral and has to stay there.

editing wavetables on the XT:
A computer can handle this better (fully agreed). But maybe this is an idea for
additional tweaking: As you choose a wavetable the wavetable is transferred from RAM/ROM to the
workmemory of the MW. Imagine that during this transfer you could do some
extra transforms on the wavetable for example :

A 16 band harmonic filter with a very simple envelope for every harmonic:
one rate(positive or negative) for the increment vs wavenumber. This would
mean just about 4 screens, would be ok as interface. No realtime control,
just as fast as switching wavetables would be ok for me.

For classic wavetables this effect can be done with SD or Philip Pilgrim’s
software, but if you put it in the box then :
– you don’t need a computer
– Input wavetables can also be the speech or algorithmic tables

Now the dreamers corner: some nice grains for granular synthesis via wavetables or some other possible way..? (the Q still has some free algos…)

>I guess it is certainly possible (I can do it on my DSS1 with a

..I support you strongly. Not that I’d have much against creating waves (or
even the worst of tsunamis :-) with the XT but not at the cost of seeing
other features left out that can only be done by the XT itself. Like a
better distortion algorithm (the distortion is already quite ok. for a
digital alg. but it still sounds mucho better if I use my stomp boxes – so
these algorithms are a waste of OS-power/space – ofcourse that is onlee my
opinion).

Sorry, but there have been no ‚real numbers‘ on this machine implemented
(except for some minor exclusions like the delay time in the fx section).
It is also quite diffucult to do so because LFO speed is stored in one byte
wheras real tempo adjustment would require more storage (eg. one byte for
the tempo in bpm and one for the factor the speed should be based on (1/4,
1/8, trioles etc). You will also face this problem, when it comes to the
envelopes and you want for instance open/close a filter in sync with your
piece of music or scan a wavetable synced to the music. So you will
(unfortunately) have to use your ears :-) Normally you can come quite
close, and – though I would rather like to have ‚real world numbers‘ – I
can live with it quite pleasantly. BTW, I heard that the Q will have hard
numbers for these time based things as well as for some other parameters.

ever thought of an envelope follower for the external input?

more about the external input:

you need more for one sound? why not put the output back to the ext.input again and put some additional filters on it?

how do you use it -purely!? easy:

.. see the XT L&R audio_in show up as independent
modulation sources in the matrix.

Something tells me that being able to modulate WaveStart with a screaming
guitar lead might lend itself to some interesting possibilities… Marzzz?

Or… Take RIGHT_IN and run it through the filters, take a sidestick track
on LEFT_IN and use it as a source for filter cutoff or resonance.

Or… Take RIGHT_IN and modulate Attack time with it, take LEFT_IN and
modulate decay time with it. Jam a sequenced pattern in through MIDI_IN…
You can now control the dynamics of the sequence by playing off of the two
audio sources…

Speaking of the arpeggiator…
I think it would be nice if an extra parameter was added….gate time.This would represent key down duration. It is very useful musically….food for thought.

Hope you manage to put it also as target in the modmatrix…..

We had a discussion on this about a year ago : Wolfram said at that time that there was not enough memory for this.

This is only true for a step based gate time. But maybe we can introduce a global gate time so that you can shorten or lengthen the whole arp rhythm. This is already very nice (we have it in the Q ;-)).

Q: The drum synth isn’t implemented but instead of that, we gave the Q a 16 part multi mode. Most of the time this is enough to do drums on e.g. 8 parts where you still have 8 parts left for other instruments. I am not sure when we implement the drum synth but I hope to see it at some time.

I vote for some basic UPAW editing (let say the FM stuff with some basic envelopes) for Onboard editing.

> >Ehem… The Wave has no UPAW at all. AFAIR.
>

There is a nice idea about randomization:

lets generate some sounds randomly and decide a „generation“ to survive.. and you should be able to choose that eg: the OSC or some settings will not be changed.. so you´ll choose what will be randomly generated and „good ideas“ (greetinx from Darwin?) will able to have „children“ and you then choose good ones.. the machine can now determine more and more what a „good mutations“ is..

and: two sounds should be „mixed/morphed“ by interpolating like this: choose 2 sounds.. press interpolate/morph.. and you´ll have a nice mixture!!

this idea is from Richard:

Mutate: the value of each parameter of each child is randomly chosen from
within the range between each parents parameter. For example if the mother
s cutoff frequency was 10 and the fathers was 88; the childrens values
would be between 10 and 88.50, 25, 88, 70, 63 etc

Mix: each parameter of each child is randomly chosen from either a parameter
from the mother or father source. For example if the mothers cutoff
frequency was 10 and the fathers was 88; the childrens values would only
be 10 or 88.

In Philip’s brilliant software (see above!), it’s just a random cross. As you
inadvertently suggest, though, it’d be cool if the synthesist could
‚vote‘ on which variations live and which die. :) Talk about
evolving pads…

from Brett: I was reading Brian Eno’s journal (A Year with Swollen Appendices),
and in it, he tosses out a great idea for a synth programming tool
that would be ideal for a machine with a great many parameters like
the uWave. (It might work even better for an even more complex machine
like the K5000s, but I’ll try to stay on topic.)

The basic idea is to apply the idea of natural selection to patch
development. Here’s how it would work: you first generate a set of 10
random patches (I’ll stick with 10, but it could be any number). You
pick one you like, and then the software generates 10 more patches
based on it by randomly altering the „parent’s“ parameters. Lather,
rinse, repeat.

In the end, you would end up ‚breeding‘ a set of patches for certain
characteristics. You could even end up with patch ‚lineages‘ that
might work well together – a track could feature a whole family tree
of patches, for instance. Another possible application would be (ahem)
sexual reproduction: the software takes two patches, and recombines
their parameters in random ways.

Eno’s explication was, as usual, terse and sketchy (he throws away
better ideas than I’ll probably ever have), but I do have a few
notions of how such a program could work. For one thing, you’d want to
be able to adjust the ‚mutation‘ amount; let’s say a master knob that
controls the maximum percentage any parameter can be altered. Even
better, mutation could be allocated according to parameter groups.
Let’s say you’ve got a patch with a great tone, but you don’t like the
way it’s filtered. So, set the Ocs Group mutation to 2% and the Filter
Group mutation to 25% before you hit the ‚Next Generation‘ button.

In the end, it’s all just a bunch of functions sitting on top of a
random patch generator. I have no idea how interesting this might be
to anyone else, but the ‚random‘ function is one of my favorite tools
to play with. I have certain programming habits and it’s always great
to discover that the machine has areas you haven’t explored yet. So a
‚genetic‘ programmer would really be a way of exploring the synth’s
design space in a non-random way.

Philip wrote:

I wrote: So this machine would be able to model the SH (yes I programmed most of
my Ju6 and ESQ1 (ESQ is not easy to do!!…) so I believe that!!!

but more than a 2 digit number and some reduced amount of knobs (if it
should sell for a LOW price –> PULSElike??

but you are right – a good idea (I would call it the Emu/Ensoniq-technique)..
for waldorf lowpower users…? (A new group?) perhaps we can get more Qs
and XTs from that money they would earn??
(you ll put it into research don´t you!!!??)
not a bad idea? wolfram? (a chance for driving a big black mercedes (CD))

but mono?? please no more monosynths! aren´t there n.off???

  <strong>UPAW in Sounddiver:</strong> perhaps you can look at an old Geerdes (BASIC/slow but easy for the

user) prog for the Atari.. the UPAW section was easier to handle as a
user… you have a lot to do before you can hear(see first) the
wanted..!!!
some things to think of is that infinite list…

this question is asked frequently: its name is Emagic Sounddiver..
use the last 2.0.6 version and have fun!!
it cool but not perfect with UPAW programming…
(there was a program from Geerdes (Atari) written in Basic
that was slower but the UPAW could be edited in a more easy way)

>I have a problem with my microwave 1(Software 2.0).

The Microwave ONE doesn’t sync to MIDI clock. It can only use MIDI
Clock as a relative source for modulations. The center (in other words: 0)
is 120 bpm while higher tempos give a positive modulation and lower tempos
give a negative modulation.
When you are searching for a „hard wired“ LFO sync, you have to go for a
Waldorf Pulse, Microwave II or (soon) the Q.
But, there is a light at the end of the tunnel: In your sequencer, just build a controller ramp with i.e. the ModWheel of i.e. one bar with the shape you need, i.e. saw or triangle. Copy this controller ramp to as many bars as you need to have it. Then use this controller to modulate what you need, i.e. cutoff. Works perfect.

the naming thing: yes – I dont have any synth that has a good name input feat. it simply take too much time.. maybe the TG77 has got a nice solution it’s the fastest of the slow: you’ve got the number pad with 3 letter.. press as long as you get one of the 3 or the number.. this is the quickest and I’d like to see it on other synths..
I’m sorry but the XT ’s (and most af all synth’s like the Z1) naming procedure is a bit silly (like tuning.. try’n to search the short wave radiostation in timbuktu)

buttons are better the knobs in this case!! The BEST one is: using / attach a simple PC Keyboard as on the EMU Samplers.. planned: that the sound name can be inserted by pressing keys on the keyboard where the low „a“ is the alphabetic start point. Similar as on the Wave.

Here’s something that is confused on the Q sometimes:
Right now, you can put unique effects on 4 of the 16 voices. If you put effects on voice 1.1, all other voices in that bank will have the same effects (1.2, 1.3 and 1.4). Same with banks 2.x, 3.x and 4.x. I only have 3 banks (A001-A100 / B001-B100 / C001-C100) – or am I understanding you incorrectly??

Yes… sorry about the confusing. When i talk about „banks“ I mean the 4 instrument keys above the display. Maybe „instrument“ is a better term.
We call them blocks to avoid confusion. 4 blocks of each 4 instruments.You switch blocks by pressing Shift+Inst.# You switch instruments in a block by pressing Inst.#
why I’ve written this?: The next OS will have effect bus routing and output support.

AND: it will send/receive CCs .. they will be routed quite intelligent: you can remote some Waldorf Synths (XT,Pulse) by simply turning the same parameter on the Q’s panel..

I had to fool around to find out this:

But : It is possible. Available FM Sources:
* Osc 1 – 3
* Noise !!!!!! here it is!!!!
* Audio In (under development) this will be great!!!!
* + other modulation sources like LFOs etc.

this was the beginning of my idea emulating the Nord Lead 2’s „Synched Noise“…

______________________________________________________
wolfram you perhaps should put a VERY LARGE FAQ button on the waldorf site (on TOP)!!! including the voices thing etc..

nice little things:

-an indicator for input level (Xternal in)
-You can’t disable program change reception on the XT.
Wouldn’t it be nice if you are playing a multi with layered sounds that the knobs would transmit not absolute values but just increment info. Example :
Sound 1 : cutoff : 80
Sound 2 : cutoff : 40
Turn the cutoff knob in ‚increment mode ‚ :
Sound1 : 81
Sound 2: 41
Anyone faked this with Logic ??? In short: a multi -relative- edit!!

this Idea should not be forgotten..
OS 1 is a 8 part multitimberal (regular XT)
OS 2 is a 1 part multitimberal (SuperWave) with 4 wavetables 8 osc, 8
lfos, 64 line mod matrix …etc
OS 3 is a 1 part multitimberal (Analog) with 6 analog emulation
oscillators, analogue emulation filters, 64 line mod matrix …etc
OS 4 is a 1 part multitimberal (FM synthesis) with 12 operators
OS 5 is a 1 part multitimberal (additive) with 32 oscillators
OS 6 is a 1 part multitimberal with Sampling abilities
OS 7 is an effects box for processing audio
All of these could be realized with a simple midi dump of the unique OS.mid :)
but it should be in one OS..!! (did somebody say Clavia?).. ;-)
A keyboard usually has no MIDI In LED. But Waldorf plan to have a „info“ display with bar graphs where you can view the activity for all 16 instruments at once (Q!!).

a Knob Box mode (XT,Q) would be nice: „somekind“ of Local Off: you can turn them.. and it sends CC (Continous Midi Controllers) to Midi but has no effect on the XT in multimode.. some workaround: choose a „null patch“ as instrument 8 (don’t say you always use all 8!!).. always set it into multi that you can select the instrument / do it via your sequencer..
Currently there is no panel „local off“ mode. I hope it will appear one day. I also hope all the knobs will transmit MIDI CC too.

Another wish: to work out how to send patch „recall“ sysex string. Much of my stuff uses Control Changes and each time I restart the song, I must manually Recall the patch as it should be before it starts changing. Anyone make the sysex for this work? I attempted to record the results of doing it but no joy… preferable as a CC!

Shift-Utility, set sequencer into record (where you want the patch saved) and hit Shift-Utility again. This saves the sysex string for the patch independent of any patch location in the MW. Will reload into the edit buffer. Is this what you were looking for!

Any other Q users able to confirm that pan is reversed on the Q (OS v1.08)?

Possible. Will be corrected with the next big update.

• How do I assign the keyboard to several MIDI channels at once – i.e. for layering and or cross-fading via keytrack?

Will be available with the soon-to-come Multi Edit.

how to do a legato mono patch. Env Mode (Shift-Release) gives the „Not yet available“ message – that’s where I’d expect retriggering options – and there’s no mono insttrument setting that I can find.

You found the right control. This is also an important issue of the next
update.

• Waldorf, please consider adding some tiny visual indicator to the LCD display to indicate a difference between a current setting for a parameter and what that setting stored in the patch is (i.e. not just the Compare which does all parameter changes). I got very used to the DW-8000 using the decimal point to indicate that a parameter value was different to the one stored in the patch – if the tweak doesn’t do what you want, you just move the slider until the dot disappears to restore that parameter setting.

Is planned but has a lower priority at the moment. plus the „where to store / is it „free“ organization: maybe it’s fine to have compare/recall and some editbuffers… like the XT..

• Waldorf, I’d find it really useful to be able to see (e.g. via scrollable LCD menu) a list of the destinations that a particular source is modulating. It’d help me to work out if an LFO/EG/etc is free more easily. At the momoment I have two options – changing the LFO and listening or Peeking at loads of different Mod source buttons (often have to Shift to see them, too).

This is also planned. Reorganization depending on either the sources or the
destinations. But this also has a lower priority.

• Can someone remind me what plans there are for the card socket? I’m sure I saw something at some time, somewhere…

Storing sounds and, in the future, adding new synthesis modules. Even updates could be stored on cards.

50-100 sound can be stored on a card.. maybe the capacity is increasing..

Someone had mentioned that their Q had once powered up with the symptom of just hanging on the initial display, (the copyright/smiley face).

This happened to mine this morning, and I thought I’d share what I found caused it.

When I powered up the Q this morning, I happened to hold down the power switch. I don’t remember this as a conscious decision, more of a space-out since it was really early and I wanted to play. :)

When the machine powered up, the display stopped at the initial display where the serial number is shown. Nothing worked, no other panel lamps were lit. I had to unplug the machine to get it to power down.

I seem to remember Stefan or Wolfram mentioning some time ago about a fail-safe boot in the MW XT that would force the machine into a state where it’s sole purpose was to listen for an OS dump from MIDI.

I’m thinking that this is what happened to me with the Q (it’s quite repeatable) and I’m wondering if this is what the other person who had the hang saw.

I didn’t play too much with how long that the power button had to be held down for this to occur, but it seems to be possible that a solid brain fart while hitting the button would do it. :)

Let me say some things regarding boot & os discussed for the Q:
since allmost all flash-based units are might be messed during an update, i decided to split the whole Q-operating system into two parts: the BOOT and the OS. The BOOT may have a max. binary size of 64kB and contains all basic i/o functions like lowlevel midi-stuff, keyboard/panel thing and even the FLASH-functions.. (and some factory-test however) – The OS may have up to 196kB and contains all the rest…

so: updating an OS is absolutely harmless, since the update-function is located in the BOOT-part, which is not touched during an OS-update… ;) In the case OS-update crashes anyhow, simply press&hold ‚multimode‘ & ‚peek‘ buttons during power up, you will get: [Please Dump System] !

The only ~critical~ part is updating a BOOT, but this is not as often subject to be updated than an OS, isn’t it ? And: don’t be nervous about the 0.87, i started anywhere from 0.00, but for those of you: i will change it for the next time
to 1.02 – or so…
here’s something weird:

Yes, there are some glitches with program changes on the current rev. It is fully understandable when you are changing a X.1 instrument, because the effects on instruments in the same block (X.2, X,3, X.4) will change too, and you will hear that change. (not really a glitch, though).

There is another little glitch where for some reason I have not figured out, patch AXX will suddenly become BXX, or some other bank change will randomly occur. The number in that bank stays the same. I really want to know how to do this on purpose! But right now it just happens on accident, I have not been able to narrow down a cause.
ok thats not wanted.. but when I switch to another sound on my XT I sometimes like to be able to program this! (this is what you meant there in the Q,too?!)
it’s somewhere between morphing and switching the wavetables knob..

ok one wavetable per OSC would be nice.. but it’s a non realistic dream..

maybe morphing non mod matrix parameters in the XT,too??!!!! (Q planned.. soon??)

c u
conseQuence
some new things.. the Wave II (or so):
if I could ever buy a Wave (a was very near as it was new… maybe if
everything is right I’d get a Wave 2 then…)

well the Wave is a cool machine.. the new Wave should include enough DSP
power to be expanded!!!
ok.. the Wave is cool.. and analog now.. but a Wave 2 should grow overtime..
and should include all those cool XT features plus Q’s ;-)
(yes.. not always possible.. I know!!)

xmorph, 3 OSCs, fx, filter and osc fm… etc…

individual Wavetables for each of the OSCs..
samples / attacksamples.. and a good connection to the outer world.. SCSI,USB..?..
of course .. all those thing I have on my site (www.sequencer.de)..

this thing should be a „all that can be done“ machine…

maybe the Q’s stepseq should be in.. the Waves Sequencer never saw the
light so…
this is a part that must not be re-invented…

big display.. and.. as I heard some of the Waves have probs to keep
their panels up..

keep it blue (important!) ;-)

if it will work I really rethink the buy-thing then…
(I would not buy a Wave 1 now…).. it should be really superior to what
is in now..
an compatible to more audio formats…

sampling?? hmm…
modular??? ;-)

most cool… but too xpensive.. displays under every knob.. (a dream..
but a cool one.. eh?)
so it could be extended and maybe completely free..

but I guess there could be a cheaper workaround then…

rethink the Matrix edit…
and: if possible full filter knobset…
for the 2 (or more) filters and all those cool params..

in the end it will be a matter of intelligent design & prices.. true!!

hmm.. when I think about all that I get the idea of a very desirable synth..
and it should have a drum editor.. like the nord lead.. is is a cool
fast tool and very useful…

the only thing I miss in the Q and XT… I love to play all sounds on
one keyboard.. THAT EASY..

the only thing I really miss (had to sell my Nord for getting my Q!)..
please not on the Wave 2 only!!.. -> wishlist for Q & XT,too!!!

to be continued…

btw: people who can read german should have read the „trilogy“ about the
Wave in KEYBOARDS!!
it is a must!!!.. I needed years to say „ok.. it is high priced and just
be cool and you need a sampler..“..

I had to decide.. now I do not need a new sampler.. ;-)
dangerous thing for me.. then.. good for Waldorf…

some more on samplers that could be done by waldorf or ???:
Actually, I had been long pining away for such a sampler. E-mu has come the
closest. One can assign MIDI CC controls to modulate just about anything using
„cords“ (as in patch cords of an old modular analog). These virtual cords, while
a little difficult to wrap your head around at first, do allow for some fairly
amazing control (all those mentioned below except for granular). Sure, there
could definitely be some improvement, but E-mu definitely has taken a step in the
right direction. Once a preset is customized with all the cords and modulators as
desired, save it and simply plug in different sample sets.

Dan
dtm productions

„ConseQuence! www.SEQUENCER.de“ wrote:
I always wondered why no company has ever made a sampler with knobs.. that means : the normal synthesis knobs plus: start,pos,end and some more nice things!!.. ok they can be assigned sometimes.. but no more!!!..
This could be a knobbed W-sampler that could store and handle its samples as algorithms like the axcel technos or so… and: there coud be knobbed funktions to morph between samples.. and nice add. like implementing grains and the like… and: it should be modular!! THAT would be nice, eh? of course it should have some addives in ROM.. the others in RAM (n.off of it!!!) and if there is an editor: please never forget the Macs!!!
cu conseQuence
>

|||| Philip |||| wrote:
> Why doesn’t Waldorf join up with Yamaha or Akai to mount a simple stripped down Q engine into one of their samplers? A multitimberal sampler with the ability to generate it’s own samples from an onboard VA engine seems to make sense to me and be the next step in sampler evolution. Sounds like a good way to sell Q boards as well. A Q board mounted in the sampler with an external knob box similar to the Access products would be all that is needed.
>
Ultimately, I would like to see a machine with a K5000s-like additive engine and a Waldorf VA engine. The sounds would be produced using the Waldorf VA then analysed by the additive synthesis engine and replayed. Sort of like a Synclavier. It would be really cool if the VA part was a „plug in“ so that you could use Q,Pulse,XT, Wave2000…etc and still store the sounds in the Additives‘ banks.

Waves done by whom?
The Wavetables No. 33-64 in the MW and no. 65 in the MW2 were actually made
by Stefan Stenzel, except the „chorus 2“ -> was done by Wolfram Franke

ever seen an open Pulse?? Philip does a modifications that add a lot of nice features:
http://members.xoom.com/NovaMod/pulse/PulseWarrantyVoid.gif
indy osc outs
Res and cutoff CV out
Osc FM of VCF with source switch and amount
Blue coloured LEDs and Different coloured 7seg digplays
Custom Start up message
User sounds loaded as presets in EPROM

see here about the mods & specs: http://www.robotnik.com/the_lab
the speech robot demos (an atari sT prog for the MW1 (microwave1):

About Speech Synthesis with the „old“ MicroWave read here:
http://www.mysunrise.ch/users/georg.mueller/waldorf/mwdemo.html

The Atari ST is saying „Resistance Is Futile“:
http://www.mysunrise.ch/users/georg.mueller/waldorf/09_st_14.mp3

Now the MicroWave:
http://www.mysunrise.ch/users/georg.mueller/waldorf/10_14.mp3

A full blown sound:
http://www.mysunrise.ch/users/georg.mueller/waldorf/20_resistance.mp3
That´s why the speech robot uses not all of the wavesamples. See Georg Müllers talk
held at PPG2002 (page 10):
http://www.mysunrise.ch/users/georg.mueller/waldorf/speech_robot.pdf