CYBERYOGI =CO=Windler

Is there battery leak acid spilled on the PCB? This can make the PCB material itself conductive and cause all kinds of trouble. Also water trapped under rubber contacts (e.g. when rained on) can make such problems. If battery acid (with alkaline batteries it is a strong lye instead of acid), everything needs to be taken apart and cleaned with water and dishwashing detergent to prevent corrosion damage. Residues than can be removed with isopropanol. Be careful with printed carbon traces on PCBs (trace bridges and key/button contacts), those tend to corrode away by acid but also might get damaged by isopropanol when scrubbing too strong.

I just started to experimented with an RTL SDR (software defined radio receiver based on a cheap TV stick - originally bought to examine EMF of sound toy chips through TEMPEST algorithms). Funny is how many strage powerful "analogue" noise timbres occur by tuning into AM on an empty (or FM) radio channel and reducing the reception bandwidth (dragging by mouse). That is really Star-Trek (original series) analogue synth stuff despite the hardware and filtering is digital.

Classic PT-series has the same sound engine like VL-1, i.e. one multipulse squarewave with linear volume envelope and LFO. It may be that the idea of synthesizing timbres this way originated from Walsh (there was that Allen Organ patent lawsuit), but multipulse squarewave is not identical with Walsh. Layering multiple Walsh drawbars would still sum them with different volume levels, while within a multipulse (16 bit steps those each can be hi or lo) the steps always keep the same height (i.e. 1-bit signal of repeating 16 steps). I remember I read in a magazine(?) long ago an interview that Casio had invented multipulse squarewave coincidentally when they experimented with LCD control voltages for calculators (consisting of blocky 4 step (2-bit) waveforms) and noticed that they sounded like organ tones. I also doubt that consonant-vowel or SD internally layers actual Walsh drawbars (which are a series of certain mathematically well defined special multipulses). It simply crossfades/morphs between 2 blocky waveforms (and than layers 2 or 4 of them having different preset analogue filters) to synthesize timbres. The internal basic waveforms were likely tweaked by ear and not planned to implement Walsh. Walsh makes only sense when you want to be capable to systematically approximate all repeating waveforms by summing a high number of multipulses in the same manner like conventional drawbars sum sines (implementing the Fourier series).

SD = "Spectrum Dynamics" synthesis. Basically it morphs between 2 blocky digital waveforms (stairwaves) and the output is additionally routed through an analogue VCF (voltage controlled filter) to change timbre. Except in HT-6000, there is only one VCF shared for all polyphony channels, so timbre of held notes changes when an additional note is played.

I have here the service manual of Casio CPS-201 (APR. 1986, paper original). CPU = "uPD7811G-226" 2x sound chip = "uPD932G" percussion IC = "HD61701" ROM = "uPD23C128EC-029" Thus it is definitely a CT-6000 variant (consonant-vowel missing link). The service manual looks quite detailed with some logic tables about filter controls, DAC signals etc.

By the way, I have added some PCB photos of CTK-1000 in this thread:

In rumbly toy-grade keyboards I also often used adhesive window insulation foam rubber strips in various spots. Use the slightly more expensive silicone-based version, because cheap (PU based) foam rubber tends to crumble apart 1.) after a decade by ozone and 2.) by battery leak vapour (and possibly also less acrid things like oil smell) which makes a big mess.

Long ago I bought this thing and I have the manual as PDF (3.2MB). Be aware that this piece of poop is not a rompler. It contains no sound generator at all but only can read midi files from diskette and send it through midi-out to an external midi keyboard.

Casio CPS-201 is AFAIK a castrated CT-6000, which uses a special variant of consonant-vowel (i.e. multiple stairwaves layered and routed through switchable fixed filters).

Casio models with and without velocity generally used different software and velocity needs more IC pins, hence velocity was not a simple feature (like a key matrix easteregg) to turn on in more expensive versions but a fundamental change in soft- and hardware architecture. Beside CTK-1000 (which shares at least *some* of the "PCM" sound) I am not aware of anything with classic PCM/CD sound set and velocity. The only keyboard with velocity I am aware of having twins without is the much newer Yamaha PSS-A50 (awesome mini keyboard) with its simpler variants PSS-F30 and PSS-E30. But despite 99% identical hardware, also here the software strongly differs (completely changed user interface and count of preset sounds etc.) and only the A50 (with velocity) has the USB midi chip, thus swapping the rom (tiny serial flash thing that tends to die easily by power glitches) likely can not turn E30/F30 into A50. The F30 firmware may be identical with the fullsize F50 but has both stereo channels wired to a mono amp, i.e. possibly there is stereo sound also inside the A50, but I haven't investigated this. http://sandsoftwaresound.net/yamaha-pss-a50-look-inside/ I don't know if such things now also exist in very new Casios, but models until 1990th definitely had different soft- & hardware.

Most strange is, the CTK-1000 hardware looks like when it had been intended to become something else (a real synthesizer??) with completely different control panel. Look at the right side of the panel PCB with empty section and cable mess. It is wired to a daughterboard with external MCU NEC D78CP14CW (shown to the left of the mainboard) that has unused ribbon cable solder pads (kind of generic prototype adapter?, or for an omitted LCD or such things?) that may translate the input data to the main CPU. It would be exciting to analyze what it does and if it dumbs down a much more versatile serious synth that never got finished. (I have no time at the moment.) (I hope I didn't clog the forum too much with almost fullres PCB pictures. I already reduced resolution but tried to keep most of the traces visible.)

The resynthesis approach resembles CELP but is likely more complex (neural nets reminds to Hartmann Neuron synth). Did you know that Casio already made 2 singing keyboards decades ago? Namely there was the Casio SK-60 (toy sampling keyboard with lofi doo-whop samples etc.) and there was the CT-840 which can sing musical notes (Do,Re,Me...) in a female lofi sampled voice.

Traditionally Casio used negative center power supplies, so many 1980th Casio keyboards got killed by plugging a wrong polarity (positive center) Yamaha or AC PSU in. (This happened very often on shopping center demo stands.) Only after Casio made fully digital PCM keyboards, they added a protection diode to prevent damage. To make it worse, my much newer Casio SA-76 now has a positive center PSU (i.e. compatible with Yamaha) and so won't work with classic Casio adapters. Most generic wall wart power supplies have a switch or reversible plug (the end at the PSU) to select polarity and voltage. Ampere rating of a PSU needs to be same or higher than the device to be powered. Voltage should be the same (slightly lower may work in devices needing >5V). Most home keyboards only draw full current at maximum volume, thus at ambient volume typical 2A keyboards may work well with 1A or less. But better avoid this with modern keyboards containing writable flash memory (XW series etc.), which may crash when voltage drops too low during write access (in worst case committing suicide by overwriting its firmware). Be careful with old unregulated PSU (heavy by their iron transformer). They output higher voltage than rated (e.g. 18V instead of 12V) when the device draws less than the rated current, so set them at lower voltage. I e.g. set them to 6V instead of 9V or 7.5V instead of 12V with 1980/90th Casio keyboards.

Cheapo mikes are almost always electret (containing a permanent electrostatic crystal), which needs no high phantom voltage but only few volts (through a resistor) to drive the internal transistor or amp chip. Such microphones often even come with 5€ soundtoys or built into all kind of cheap recording devices. Due to internal amplification they output a fairly high signal level. https://en.wikipedia.org/wiki/Electret_microphone Dynamic microphones are physically bigger and often more expensive (but can have better bass response), thus inside laptops, small cameras etc. normally only electret mics are used. Also those "USB" mics (in headsets, desktop accessories etc.) are normally electret. AFAIK modern devices like smartspeakers (Alexa) even contain an array of a dozen of those tiny mics to cancel ambient noises and focus the listening direction to the talking user. Before there was electret, cheapo mics often were piezo (dielectric crystal attached to aluminium foil diagphragm), those tended to sound hissy but also had high signal level. Nowadays they are mostly used as contact mics (e.g. pickups for nylon guitar strings or burglar alarms), those are the same component like the very flat piezo speakers inside 1980th LCD games etc. https://en.wikipedia.org/wiki/Contact_microphone

I once repaired a Casio PT-50 that was soaked in battery acid, so its chord button PCB apparently had an acid puddle under the rubber contacts and lost most though-hole contacts and carbon paint. The conductive carbon paint is mainly for key contacts themselves (so they won't oxidize), but also adhesive or glued copper foil (tinned with solder) may work sufficiently. Use thin enameled wire for the rest. I used conductive silver paint (had no carbon) which may corrode over time, but the center piece of most carbon contacts was still there and mainly the metal trace was gone. I don't know how even and reliable the key response will be (CTK-1000 has velocity), but patching such things together is possible. See here. There are plenty of black rotten spots on your traces, so I recommend to place the wires in different paths than the original physical layout. The logical layouts (seems to be groups of 8 interconnected by diodes) may be easier to follow by making the insulated wires cross each others.

Clean the rusty contacts with isopropanol and patch broken traces (watch for tiny black spots and use a multimeter) by soldering coil wires acros it. Conductive carbon paint can be cheaply ordered on eBay (from China). Its tiny plastic container has short shelf life and tends to dry out soon. To reuse it, you will need to add new solvent.

When polyphony is higher than 2, the model does not matter. But the general layout (count of rows x columns) matters (most Casios use 6er groups) and if it has velocity (CTK-1000 does). If a PCB got massively damaged by battery "acid" (which is actually a lye in alkaline batteries) the epoxy may swell and become conductive. I read that soaking the spots in pure liquid dishwashing detergent for a day (cover it to prevent drying out) can help to neutralize the chemical. (I did this with an LCD watch.) Then rinse it with freshwater an towel it dry. (You may also use a hairdryer at low heat.) I have occasionally cleaned Casio PCB and rubber contact parts with a cotton swab and isopropanol, but do not scrub hard on those carbon traces. Also the carbon track in some potentiometers (e.g. 1980th Bontempi slide pots) can get damaged by isopropanol. Particularly NEVER spray "contact cleaner" fluid into them, which random solvent mixture is way too aggressive. Otherwise I got a very dirty Casio MT-70 which crackled like hell and made no sound in most potentiometer positions. Normally I would take the pots apart and clean inside, but in this model this needs to unscrew and remove half a dozen of stacked PCBs, which even is risky due to the flimsy LCD foil cable. So I gave it a try to slowly pry off the rotary knobs by screwdriver and pour isopropanol (better use 99%/waterfree one instead of generic 70%) down the shaft from the panel side. To make it flow down, I pushed an airpump hose around each shaft and pressed the airmat bellow to force the liquid into it. After playing decathlon on each of them (i.e. turning about 30 times in both directions while listening) the pots worked almost like new.

Stephen Hawking's speech software was much more complex (and driving a hardware speech chip). It was Hawking himself who programmed the parameter set to customize his remade voice. He was fairly envy about its voice and eventually did not want to replace it with more natural sounding newer speech synths because he considered it part of his cyborg personality. (Nowadays liarbird technology would have been used for such things.) https://www.scienceabc.com/innovation/stephen-hawking-cheek-communication-help-computer-speech-generating-device.html https://www.wired.com/2015/01/intel-gave-stephen-hawking-voice/ This one may be fake: https://lingojam.com/StephenHawkingVoiceGenerator SAM can not even do pitchbend for pronunciation and sounds very husky. (I know the C64 version, which needs no additional hardware and also runs on emulators. AFAIK "Tales of Arabian Nights" uses it - listen to youtube examples.) I grew up with Amiga (owning an A500 modded into a huge steampunk-like wooden desktop case). The Amiga speech is pure software (Narrator library) and can control pitch and tempo per syllable for pronunciation. AFAIK source code of both have been found. Amiga speech can sing, but tended to make pauses in unwanted spots because the 68000 CPU was too slow to render a continuous wave output fast enough. The commercial speech synth which timbre IMO came closest to Hawking's Voice is my Hexaglot EG-6000 (a talking organizer with optional translation cartridges). I also own a Hexaglot Square One with similar voice (only through headphone jack, LCD is messed up by broken foil cable). Also these have a general pitch and tempo control.

Casio PT-100 hardware has tons of eastereggs. I only absolutely hate this crappy fartsmone-ready $*@%! forum GUI that does not let me properly edit anything despite Firefox pastes HTML. So the bullets jam and spill dud jackets everywhere. I haven't tried much with MT-520 although I got the original service manual. An MT-28 variant without ROM-Pack was released as Casio MT-25; instead of the "guide select" slide switch it apparently has a 2 step "casio chord" switch for single finger accompaniment like PT-100. Also MT-28 and PT-180 are crippled variants. A longer 44 midsize keys version of the PT-100 was released as Casio MT-105 (aka PT-200? | 2 speakers, stereo?) and a mono version as MT-56; an MT-56 variant with sequencer and each 12 sounds and rhythms (buttons like MT-28) came out as MT-55 (all seen on eBay). They all have blip percussion. (With this hardware class the chord section split point on the case shows the percussion type; only melody section at F#2 means analogue.) keyboard matrix 83 KO1 84 KO2 85 KO3 86 KO4 87 KO5 88 KO6 89 KO7 90 KO8 91 KO9 92 KO10 93 KO11 94 KO12 pin out 1 out 2 out 3 out 4 out 5 out 6 out 7 out 8 out 9 out 10 out 11 out 12 out/ in demo memory start/ stop R. synchro/ key select R. key select R. tempo - R. stop one key play 2 sustain off G. IV model select 1 [diode] model select 2 [diode] in 8 47 KI8 memory record R. synchro R. fill-in R. tempo + R. start/ stop one key play 1 sustain on G. II R. bank select O. bank select in 7 46 KI7 o F1 o B1 o F2 o B2 o F3 o B3 o F4 o B5 G. III M. chord guide 2 R. samba O. violin in 6 45 KI6 o E1 o A#1 o E2 o A#2 o E3 o A#3 o E4 o A#4 G. I M. chord guide R. swing 4 beat O. strings in 5 44 KI5 o D#1 o A1 o D#2 o A2 o D#3 o A3 o D#4 o A4 C. casio chord M. melody guide R. swing 2 beat O. flute in 4 43 KI4 o D1 o G#1 o D2 o G#2 o D3 o G#3 o D4 o G#4 C. fingered chord M. auto play R. 16 beat O. pipe organ in 3 42 KI3 o C#1 o G1 o C#2 o G2 o C#3 o G3 o C#4 o G4 C. chord off M play R. disco O. harpsichord in 2 41 KI2 o C1 o F#1 o C2 o F#2 o C3 o F#3 o C4 o F#4 o C5 M. power off R. rock O. piano in 1 40 KI1 All unknown function names and in/ out numbers in this chart were chosen by me. The input lines are active- high, i.e. react on +Vs. Any functions can be triggered by a non- locking switch in series to a diode from one "out" to one "in" pin. In MT-88 the key leds are wired through 12 transistors to CPU pins 66..71 and 73..78, multiplexed though pin 72 and 79 those switch 2 ICs 74HC174P. legend: "o" = keyboard key underlined = function needs locking switch (i.e. stays active only so long the switch is closed) R. = rhythm C. = chord O. = orchestra (main voice sound) M. = 'mode' switch G. = 'melody guide' switch orange background = easteregg (unconnected feature) Fist I had made very incomplete schematics of PT-100. Later I studied the MT-56 and MT-105 service manuals and analyzed the MT-88 to complete it. In PT-100 I had measured a 'rhythm stop' doublet at KO3->KI7, but it was likely a mistake because at least in MT-88 it does not exist. Also 'C. chord off' and 'C. fingered' were swapped. In MT-56 service manual there is a fixed diode shown at KO11->KI8 with the hint "model selection MT-56" and "(MT-56 selection)"; this likely switches the CPU (key matrix behaviour?) into a different mode (octave low?). main voice sounds ("tone") The main voice sounds are selected by 6 individual button inputs together with a bank switch input. Normally these are controlled by the 8 position "tone" slide switch which automatically activates the bank switch through a 2nd contact row while selecting the sounds {vibraphone, jazz organ, synth sound}. To get access to all 12 preset sounds, disconnect the "O. bank select" line from the slide switch and instead wire it through a diode to a (non- locking) button switch. According to comparison with other keyboards of the same hardware class, the names of the new preset sounds seem to include "violin", "clarinet" and "celesta". The last sound is a harsh electronic organ tone with fast attack phase and chorus, that resembles "brass ens." on small Casio ToneBank instruments. note: After the modification the "tone" switch positions {vibraphone, jazz organ, synth sound} will instead select {piano, pipe organ, violin} so far the 2nd bank is not selected. To select sounds will now become a little tricky, because to select any sound of the bank 2: hold down the bank button, then move the "tone" slider to the intended position, then move it to an intermediate switch position next to it, and finally release the bank button. If you release the button while the "tone" switch is still in a valid (not intermediate) position, it always switches back to bank 1. It would be likely better to add a locking switch instead of the button, but I discovered that the button can be also well used as a realtime sound control to rapidly bounce back and forward between both banks during held notes (which also re- triggers their envelope); if you are in doubt, add both wired parallel. (The "tone" slider itself responds rather slowly and thus constitutes no good OBS realtime control.) rhythms The rhythm bank select input "R. bank select" works very similarly like the sound one (see above), but fortunately it is not necessary to rewire here anything, because you can instead add the "R. key select" button that selects all 12 rhythms through the leftmost white piano keys. These rhythms are {rock, disco, 16 beat, swing 2 beat, swing 4 beat, samba, bossa nova, beguine, tango, march, slow rock, waltz} and exactly correspond to the rhythm set of small ROM-Pack keyboards (like Casio PT-82). fingered chord Originally the instrument has the "casio chord" 2 step slide switch to switch chord/ accompaniment either "off" or to single finger chord mode. But additionally also a fingered chord mode exists, which permits much more versatile accompaniment play and makes great organ basses with rhythm off. Thus I disconnected the "casio chord" switch and added instead 3 buttons to the inputs "C. chord off", "C. fingered chord" and "C. casio chord". Theoretically also a 3 step switch can be added here, but the only disadvantage of separate buttons is that the instrument always comes up in single finger mode after switching it on. sustain Either a locking switch or 2 buttons can be added to switch sustain for the main voice on and off. I rewired the disabled "casio chord" slide switch for this purpose. The pedal sustain of CT-805 is no additional easteregg but only activates normal sustain (does nothing when already on). fill-in A fill-in button can be installed at KO4->KI7. blip percussion, chord split point, octave During power-on, the CPU scans presence of these 2 fixed matrix diodes to set the general operating mode of the instrument. The PT-100 contains none, the MT-88 both. Wire them through 2 switches to get interesting behaviour changes. Theoretically even push buttons are sufficient instead of locking switches, but you would need to hold them during every power-on to select the mode. model select 2 model select 1 percussion type chord spit point octave blip A#2 hi X blip A#2 lo X blip C2 lo X X analogue F#2 lo While upgrading blip percussion instruments with analogue percussion is not worth the effort (plenty of additional discrete circuitry), the opposite is very easy and even automatically disables analogue percussion (CPU pin 98 FC4) when a blip percussion mode is selected. Great is that the chord section key split point printed on the case allows to identify which type of percussion is inside; only melody section starting on F#2 means analogue. The analogue percussion mode disables blip percussion and instead outputs an additional continuous bass voice in accompaniment and percussion trigger pulses on key matrix outputs KO1..KO7, those can be demultiplexed by AND comparison with CPU pin 98 FC4 to trigger external analogue percussion circuits (see MT-88). All blip percussion modes turn the "synchro start" button into a "synchro/ key select" button that first waits for pressing a white key to select a rhythm (and exits with wrong key press). Interesting how other modes introduce various bizarre glitches and throw a wrench into key lighting modes (of MT-88). Without these diodes, the keyboard is assigned 1 octave higher, so in melody guide mode you have to play 1 octave lower(!) than what the leds indicate, and because soon you end up in the chord section, you can not reach some requested notes at all but only step further through the song by pressing "one key play". Also the highest keyboard octave in 'flute', 'strings' and 'celesta' makes mess; above C4 they play disharmonic low notes those form no regular tone scale and grunt lower than the normal C1. Maybe a lookup table ended too soon to save ROM space, but possibly this was a test mode or even a hidden message in the chip. The keys from C#4 to C5 play the notes {B#-1, B#-1, B#-1, B#-1, C#-1, C#-1, C#-1, B#0, B#0, G1, E#1, B-1}. At least the G1 and E#1 are detuned against the normal notes. The falling sequence length of equal notes hints to a part of a waveform or envelope curve, or did aliens compose this sequence to contact us?!? (Recording this in the MT-88 sequencer and replaying it on a normal preset sound plays the regular notes C#4 to C5, which proves that the glitch is in the sound engine itself and not keyboard decoding. Switching back to normal diodes keeps the high and faulty notes in the sequencer, which shows that the SRAM stores note numbers and not key numbers.) With "model select 2" diode the chord section is too short to play chords like displayed in chord guide mode. Instead you have to play those chord notes an octave lower, which is pretty confusing. Only with "model select 1" or both diodes the guide modes work ok. When I first saw that Casio MT-88 used the same CPU like PT-100, it remained a great riddle to me because they sounded and behaved so different. I expected complex tricks like an external ROM and that the same digital percussion output was gated and routed through a variety of external filters to change their timbres. When I found out that the only other digital IC was SRAM and not ROM and no digital percussion waveform was coming from the CPU, I got even more puzzled until I finally saw in the MT-56 service manual a fixed matrix diode KO11->KI8 "MT-56 Selection" (also in MT-105 schematics). So I analyzed my MT-88 and found even 2 those change behaviour when absent. rom pack modes The additional 'melody guide' and 'mode' switch settings need a rom pack port. The 'chord guide 2' mode not even exists in MT-88; it combines chord guide for the left hand with free play (key leds off, no automatic melody) for the right hand. Installing a ROM-Pack port to the CPU would be electrically simple but mechanically complicated. Adding key led circuitry is not really worth the effort. separate volume controls The CPU outputs many sound channels on individual pins, thus separate volume controls can be added here. timbre filter switches Like in CT-410V, the main voice is routed through fixed timbre filters, those in MT-88 are controlled by 2 CPU outputs pins 100 FC5 (coarse) and 1 FC6 (fine) for up to 4 timbre settings. In cheaper hardware variants like PT-100 the filter uses only FC6 for 2 timbres (seen in MT-56 & MT-105 service manuals). Pin 2 FC7 is no filter control but mutes the main voice during "mode" or "chord" switch operation to avoid popping. To make the filter switchable, cut the trace at pin 100 and 1 (I did it near the CPU but you may use a safer spot, e.g. at the "4066" IC of the filter). Install a switch with at least 3 positions into each of them to input either the original CPU pin, hi (+5V) or lo (GND) to the filter. A 2 position slide switch with open (intermediate) center position will work as well when you solder a pulldown resistor (I used 22k) against GND to the filter control input and make the switch only select between CPU output and +5V. Of course you may also install potentiometers (like I did in CT-410V) to distort the timbre. In keyboards without 4066 this may even permit gradual changes instead. (In my wiring I installed a pluggable connector at the switches to improve serviceability.) preset sound: FC5 (coarse) FC6 (fine) piano H harpsichord H pipe organ H flute H strings H violin vibraphone H celesta H jazz organ H clarinet H reed synth sound H Pin FC5=hi makes the timbre duller, while FC6=hi makes it slightly brighter. The combination of both hi is originally not used, but produces an intermediate sound. others The sequencer function 'memory start/stop' is at KO2->KI8 needs an optional sequencer SRAM "NEC D446C-3"; without it will crash (lockup?).

Casio tends to use printed conductive carbon paint for key contacts and wire bridges (actually acting as resistors), those get eaten away by battery acid (and likely by vinegar and baking soda too, so be careful!). They do survive water and dishwashing detergent, so take the keys PCB out and wash it. Also wash the rubber contact strip. Wash the soap off and dry it by hairdryer before putting things back together, Where key contacts are gone, try conductive paint or simply glue a piece of copper foil with soldered wire on the PCB to fix it. Also though-hole copper contacts in the PCB often get eaten by acid and need to be replaced by a wire bridge.

Yes, MT-205 is MT-520 based. The CPU is close relative of PT-100 and MT-88 but with different software, and because MT-520 has a sample based percussion IC "OKI M6294-02", it doesn't need the CPU blip percussion channel and so uses the surplus polyphony to layer 2 audio outputs for a thicker main voice in the manner of classic Consonant-Vowel synthesis and even can do a stereo chorus within the CPU. Percussion often differs among Casio models with same CPU due to changed analogue filters and in analogue percussion instruments even omitted individual drums (output pin disconnected or wired to another drum). E.g. the famous rusty base drum of PT-100 sounds much tamer in other keyboards with this CPU. E.g. MT-40 had a terribly dull analogue snare that got fixed in later models. Casio PT-100 is a cheapened single-chip version of consonant-vowel hardware with accompaniment (sort of dumbed down MT-65). The LSI "Hitachi HD61702xxx" (100 pin SMD, pins count anticlockwise, xxx = software number of internal ROM) is the CPU of Casio's last generation of non-PCM home keyboards. It was their most sophisticated squarewave and stairwave hardware with functionality resembling a D930G and D931C pair with DAC integrated into a single-chip. Unfortunately compared with D930G the functionality is very restricted. So the keyboard matrix supports only 12 preset sounds and rhythms, and beside a simple sustain switch there are no sound or accompaniment variations. The HD61702 has separate analogue outputs for polyphonic melody, chord, bass and obligato channel; the 8 note polyphony is shared among these. Also the blip percussion (can be disabled through 2 matrix diodes) is part of bass and obligato. Like the early HD44140, the upper and lower DAC bits for each channel have separate outputs those need to be combined through an external voltage divider (small resistor network) to reduce noise; the mixing ratio is 1:69 (1:110 at both ends of bit compensation trimmer) for main voice and 1:100 for others. All DAC outputs need 390 Ohm pulldown resistors against DAC ground 0V. External trimmers can be used for bit compensation to adjust smooth transition. The internal DACs also have plenty of individual supply voltage and ground pins to permit signal tweaking, and even their LSB is accessible on pins 25..34. There are 3 switch outputs for fixed main voice filters; one is only used as a pop blocker to mute during switch operations ('mode', 'chord', 'memory stop', but not preset sound changes). The CPU supports a ROM-Pack port and key lighting LEDs. An optional SRAM can be connected as sequencer memory. The sound generator seems quite similar like D931C. The preset sound waveforms are 5 bit high (up to 30 steps during attack, 18 steps for continuous tones) and apparently consist of 2 stairwaves with independent volume envelope, those can be detuned against each others for chorus and phasing effects. Most waveforms are symmetrical and seem to be made from up to 4 quaterwaves those are each 8 steps long and can be flipped and mirrored. Like with all stairwave instruments there is no interpolation, so the wave shape and height does not change with note pitch and all blockiness also remains in bass range (unless smoothed by external fixed filters). Interesting is how the waveforms are composed. E.g. the 'flute' resembles a blocky sine with small square "nipples" on their peaks those vibrate irregularly for wind noise. The 'piano' uses a strange geometrical trapeze shape with varying sunken center to change the harmonic contents. It was really an art to construct good sounding waveforms at that time, and yes it can sound nice. However unlike D931C the envelopes seem to be linear and end too soon, but this also may be result of too low DAC resolution. As far my analogue scope can tell, it may have only 8 bit (5 bit waveforms + 3 bits for 8 polyphony channels), but possibly the envelope resolution is a little finer. The blip percussion even seems to use only one DAC halve; at least their decay envelope doesn't look linear, and lissajous patterns on my scope hint that rough timbres are not samples but partly contain shift register feedback noise with varying bit patterns. Possibly the synthesis uses noise modulation (US patent 4656428) that adds random numbers to the readout address of a very low resolution sample to make hiss timbres more vivid. The versions of "Hitachi HD61702xxx" ("xxx" = software number of internal ROM) differ in accompaniments, preset sounds and additional features. software number hardware class notes & features A02 MT-88/PT-100 original version A03 MT-520 Super Drums support, uses percussion IC M6294-02, melody=2 layered outputs A04 SK-200, SK-100 combined with sampling CPU M6283-02 + percussion IC A06 Hohner PK60 (SK-100 variant) modified A04?, combined with sampling CPU M6283-06 + percussion IC

Speak&Spell hardware has been fully deciphered and is emulated in MAME (so you can record the WAV output). Also various old synths and keyboards (including even a somewhat off-sounding Casio CTK-551) exist in MAME now, which AFAIK even supports midi-in (haven't tried). If you want the real thing, there is even a midi kit for Speak&Spell: https://hackaday.com/2012/02/09/midi-controlled-speak-and-spell Sharp speech synthesis AFAIK the Sharp speech engine has not been emulated yet. I collect and have repaired many Sharp talking clocks, (rebranded) watches and calculators. The CT-660 exists in a German (CT-660G), an English (CT-660) and a Japanese language version. Strange is that on eBay the German version is much more common than the English one, while I never saw the Japanese model at all (see youtube example, front is labelled "ELSI QUARTZ" instead of "TALKING TIME"), so it may have been a prototype. I love the way the alarm starts with a 5-note jingle, then announces the time and then plays a longer squarewave melody ending with a trill (a bit like a ringtone). When waiting 5 minutes, it repeats the alarm and says "please hurry" or something like that (German version "Bitte beeilen!"). The German version rolls the "R" in a funny way and so e.g. pronounces 11 as "Errlf" instead of "Elf". The Sharp CT-660 was the world first digital talking alarm clock and initially very expensive (200 US$ or such). It has a volume knob and some models have a little silicone rubber plug at the left case site. When poked out (be careful - the material rips easily apart) it reveals 2 pins wired parallel to the speech button (keyboard matrix), so it could be installed in contraptions to automatically announce time or act as a stop clock, or connect a bigger button for impaired people. (The yellow one on top is IMO too small for an alarm clock, and only switches to slumber instead of proper "alarm off", so it keeps repeating until using the slide switch in the lid at case bottom.) The Sharp CT-661/665 has simpler functions (no stopwatch etc.), different melody (very shortened "Sah ein Kab ein Röslein stehn", less nicely made) and operation is optimized for the blind (i.e. different sounds guide through clock set modes instead of all those slide switches). The 1980th "Vox Clock 2" contains the same hardware without LCD and has a lovely male robot voice, which despite graininess is well understandable. It only has a little speech glitch that pronounces 12 as something like "thrown" or "throne" instead of "twelve" (like when the wavetable algorithm fails to say "two" and "one" at the same time). The 1990th "Vox Clock 2" has an additional LCD but uses a sample based chip with English female voice and only 4 beeps instead of melody. The CT-660 hardware is quite complex, containing a clock CPU (32KHz) on the front PCB, a separate speech CPU (4.1MHz), a DAC (or sound?) chip and audio amplifier. Most of the PCB is occupied by a quite big discrete stepup converter to increase the 3V battery voltage (2x AA cells) to about 5.5V(?) for louder speech output. I later bought a broken specimen that instead of the external DAC chip has a hybrid resistor ladder DAC with thinner sounding voice. (Only the LCD worked no speech and some buttons failed because traces were corroded by battery and someone drowned it in oil - yuck!) The Sharp EL-640 talking calculator even has 2 speech CPUs (or an additional ROM?) because it speaks more words (for clock and calculator) and runs on 4 AA cells. The simpler EL-620 (no clock) is slimmer and so depends on 2 unusual thick button cells. Sharp voice synthesizer things use a grainy but nice sounding kind of wavetable speech hardware. Like Speak&Spell, when shitshot by power glitch (battery wiggling) they make plenty of freakish noises. E.g. the chip can playback speech at half speed or interprete the same data either as speech or musical notes (squarewave with linear decay envelope varying with note length). Even the EL-620 does this despite it has no melody. Talking watches with Sharp speech chips (e.g. by Trafalgar, Omni VoiceMaster, Micronta VoxWatch, MeisterAnker) use similar technology with smaller COB chips. Loose solder joints at SMD parts and broken PCB traces are typical issues Instead of lithium they unfortunately use strange thick alkaline button cells those are prone to leak forgotten inside. I wrote more about the Sharp speech hardware here: https://forums.bannister.org/ubbthreads.php?ubb=showflat&Number=120188#Post120188 https://forums.bannister.org/ubbthreads.php?ubb=showflat&Number=120195#Post120195 Despite many talking clock brands, only few speech engines exist. In 1980th beside Sharp only Seiko made their own speech synth chips. Seiko (WristTalk A964, A965, A966) watches use grainy lofi male samples and have a big COB containing 3 silicon dies, which has user selectable English and one other language. The Seiko world time clocks (World Time Voice Alarm DA716K) speaks English and seems to use grainy female wavetable voice. The pyramid alarm clocks (PyramidTalk) use a very different chip for each language and have female voice. English and Japan version have calendar with date and day display (LCD layouts differ). The German issue has none and the user interface differs (hold up/down buttons to set time). 1990th Seiko watches have a higher resolution female sample voice. Most other 1990th talking watches and alarm clocks use Holtek COB chips, those are sample based and can be recognized by the rooster (cockadoodledoo) alarm (often also cuckoo and some others) and typically female sample voice. Casio speech synth hardware Also Casio made very few talking alarm clocks and calculators using own SMD speech chips (grainy wavetable like Sharp, but female voice). I own the Casio SQ-200 (cube shaped LCD clock) and of course a TA-1000. I experimented a lot with the Casio TA-1000 speech CPU "Hitachi HD61912 C02, 3M13" (60 pin SMD) which communicates with a main CPU that is likely a "NEC D1864G" variant (64 pin SMD) like in Casio ML-81, ML-90 and such. I desoldered various pins and examined the behaviour. I also found test pins in Casio calculators and VL-Tone, those display strange counting numbers on LCD and output data (rom contents?) on keyboard matrix pins. What I mean with that Alexa is no fair comparison with offline speech synths is because these speech assistant cloud apps render everything online in datacenters fed with huge quantities of bigdata. So it would not surprise me if their neural network continuously compares spoken and written versions of the same texts (e.g. official TV news) from the internet to optimize pronunciation.

AFAIK the first Yamaha PortaSound keyboards used foil contacts (much like in nowadays cheap PC keyboards) instead of conductive rubber, which may have been considered too unreliable at that time. How long rubber key contacts last depends not only on the material (quality silicone vs. butadien rubber or shortlived plasticized PVC trash) but on how the key mechanism acts on the end of key travel to prevent crushing the contact (or punching a hole through) when repeatedly pushing keys too hard. I.e. the key end has to knock on a hard surface (hopfully buffered by felt to muffle the knock) before the rubber dome gets completely crushed. That's likely why good aftertouch keyboards are more expensive than only velocity sensitive ones. While velocity only measures the timing between pressing 2 such contacts under each key, aftertouch (at least polyphonic) needs to sense the actual pressure (i.a. electrical resistance) on each contact, which makes them more vulnerable to chemical decomposition or mechanical variation and wear of each key. (Hence e.g. the monophonic aftertouch in Casio CT-6000 simply senses rotation of the entire key mech metal frame by a potentiometer to avoid this.)

For robot voices controlled by keys there is the payware softsynth Chipspeech, which contains plenty of remodelled versions of historical speech synthesis chips (listen to the youtube examples). https://en.wikipedia.org/wiki/Chipspeech Alexa is no fair comparison to any PC based text2speech, because its cloud AI likely dynamically accesses many GB (or TB?) of online text examples to pronounce things right. I find classic speech synths made from a few KB of memory much more exciting. Did you know that already in 1980th Casio released a few speech synthesizer products? These included the talking clock calculator TA-1000 (female robot voice) and several talking alarm clocks, although AFAIK none of them got famous or were particularly successful. I experimented with the TA-1000 speech chip; by behaviour it is likely a crude kind of wavetable synthesis (data reduction by concatenating waveform samples) similar like that in early Sharp talking clocks and calculators but with different voice. AFAIK a predecessor of Vocaloid did use synthesized speech before memory got cheap enough to use actual singer's phoneme samples.

The Casiotone 7000 has many stereo panning modes and can save sequencer data on datasette. The main voice corresponds to MT-65 (soundchip D931C) but is here controlled through an external CPU to handle 61 keys (later 61 key models used the normal accompaniment CPU D930G with changed firmware).