haha I bought one too and dumped it. It is only half of the set though, so there are other words that are missing. The same seller also has another chip with the code SKQQ and I dumped this also. Unfortunately this is only the top half of a set, so it's impossible to decode what's on it. The first ROM has the word sample tables and with those missing, it's impossible to decode the rest of the data. The chip set in the video has 117 phrases in it, but only the first 49 are usable (and ironically the very last phrase, which is silence) because the rest point to data on the second ROM that is missing. (to figure out the word count, you can inspect the beginning of the ROM, each word has a 2 byte pointer. generally the first word's pointer will point to the end of the pointer list. In the case of this ROM, the first word points to address 00EA which means there are 0xea/2 = 117 words. divided by 2 since each pointer is 2 bytes).
Great! Thanks. Makes a LOT of sense! I have SKQQ too and it only makes noises. I did figure out that SKQQ is the slave ROM to some unknown master ROM, and that SHAO/D is a master ROM without its slave ROM. What you have found out is never discussed in any of the old data sheets. You and I appear to be some of the very few out here who are trying to unravel the secrets of how this system actually works. Someone outbid me on that board that has 54104 processor plus SSR 1,2,5,6. Was that you by chance? :)
@@leroyjones6958 yes that was me. sorry. I DID trace the schematic out, though if you want it! I dumped the four ROMs and they match the existing dumps so you can just burn the data to an EPROM/flash chip and use that. I reverse engineered the chip 100% back in 2007 and wrote an emulator for it. The datasheet is intentionally vague on how it works. It is pretty straight forward, though. each word is a list of 3 byte entries pointing to sample pieces. first byte is repeats/sample length and stop flag, next two bytes is pointer to the data, and encoding type. weirdly the word pointers are big endian and the sample pointers are little endian. this is why without the first half of the data the sample data is nearly impossible to decode. it is missing repeat count and lengths and the order they are used.
@@kevtris So how did that complete set on that board turn out? Do all of the Digitalker ICs work correctly? I am not quite well enough up to speed quite yet to be able to dump and burn my own ROMs. Close though. I do have and use a home m,ade EPROM programmer that uses parallel port and is operated by a script I wrote for it in Crosstalk Mar 4, version 2.0. That runs in DOS. Currently using an old IBM 750 P90 machine. I also want to learn enough more so that I can eventually use the venerable 600 word library for Digitalker words that is on github. At present however I am still at the stage of rounding up what existing ROMs I can find. I only have (2) sets of SSRs 5 and 6. Would like to find more. but as it ended up, you wanted that board more than I did because at the end it was getting a bit too rich for my blood. I am wanting to have a serious Digitalker collection. I use it in a telephone system as a piece of "common pool" equipment. The outstanding feature of Digitalker is its speech clarity and ease of use. Hey did you notice how poorly sounded-out the words are on that game chip? Does not even sound like Digitalker, sounds more like SPO-256. That ROM must not have been coded using Mozer's patented method. What do you think?
@@leroyjones6958 yeah all the chips seemed to work. The reason I was interested in it was to trace out the circuit on the board; it is designed to plug into an apple 2 and give it digitalker capability. The design is pretty interesting and wasn't quite finished but it was 99% of the way there. it would've worked as-is and with 1 minor modification been perfectly usable. He used JK flipflops that toggle each time it is written to, so you could change between SSR1/2 and SSR5/6 but there was no way to know which set was in use! I have a complete schematic of the board now. As for the sound quality, it indeed is pretty poor on that SHAQ chip (it's actually SHAQ and not SHAO. Q = ceramic DIP, N = plastic I think) and I think there is a reason it does not have the "mozer" name on it. It is monotone and indeed sounds like it was made out of phonemes and I too thought it sounded very sp0256 like. Speaking of, I have a big collection of SP0256 and SP0264 chips and their associated speech ROMs, and have dumped all of those as well. hit up my email and I can send you an EPROM with several digitalker sets in it if programming them is difficult for you, so long as you're in the US. you can find it on my web page. search for my youtube username and it should come right up then the contact page. I have a lot of speech synthesis related stuff up.
Feels like its part of a coin operated gambling game circa 1980's
00F sounds like "Held" to me.
Probably right.
haha I bought one too and dumped it. It is only half of the set though, so there are other words that are missing. The same seller also has another chip with the code SKQQ and I dumped this also. Unfortunately this is only the top half of a set, so it's impossible to decode what's on it. The first ROM has the word sample tables and with those missing, it's impossible to decode the rest of the data. The chip set in the video has 117 phrases in it, but only the first 49 are usable (and ironically the very last phrase, which is silence) because the rest point to data on the second ROM that is missing. (to figure out the word count, you can inspect the beginning of the ROM, each word has a 2 byte pointer. generally the first word's pointer will point to the end of the pointer list. In the case of this ROM, the first word points to address 00EA which means there are 0xea/2 = 117 words. divided by 2 since each pointer is 2 bytes).
Great! Thanks. Makes a LOT of sense! I have SKQQ too and it only makes noises. I did figure out that SKQQ is the slave ROM to some unknown master ROM, and that SHAO/D is a master ROM without its slave ROM. What you have found out
is never discussed in any of the old data sheets.
You and I appear to be some of the very few out here who are trying to unravel the secrets of how this system actually works. Someone outbid me on that board that has 54104 processor plus SSR 1,2,5,6. Was that you by chance? :)
@@leroyjones6958 yes that was me. sorry. I DID trace the schematic out, though if you want it! I dumped the four ROMs and they match the existing dumps so you can just burn the data to an EPROM/flash chip and use that. I reverse engineered the chip 100% back in 2007 and wrote an emulator for it. The datasheet is intentionally vague on how it works. It is pretty straight forward, though. each word is a list of 3 byte entries pointing to sample pieces. first byte is repeats/sample length and stop flag, next two bytes is pointer to the data, and encoding type. weirdly the word pointers are big endian and the sample pointers are little endian. this is why without the first half of the data the sample data is nearly impossible to decode. it is missing repeat count and lengths and the order they are used.
@@kevtris So how did that complete set on that board turn out? Do all of the Digitalker ICs work correctly? I am not quite well enough up to speed quite yet to be able to dump and burn my own ROMs. Close though. I do have and use a home m,ade EPROM programmer that uses parallel port and is operated by a script I wrote for it in Crosstalk Mar 4, version 2.0. That runs in DOS. Currently using an old IBM 750 P90
machine. I also want to learn enough more so that I can eventually use the venerable 600 word library for Digitalker words that is on github. At present however I am still at the stage of rounding up what existing ROMs I can find. I only have (2) sets of SSRs 5 and 6. Would like to find more.
but as it ended up, you wanted that board
more than I did because at the end it was getting a bit too rich for my blood.
I am wanting to have a serious Digitalker collection. I use it in a telephone system
as a piece of "common pool" equipment.
The outstanding feature of Digitalker
is its speech clarity and ease of use.
Hey did you notice how poorly sounded-out the words are on that game chip? Does not even sound like Digitalker, sounds more like SPO-256. That ROM must not have been coded using Mozer's patented method. What do you think?
@@leroyjones6958 yeah all the chips seemed to work. The reason I was interested in it was to trace out the circuit on the board; it is designed to plug into an apple 2 and give it digitalker capability. The design is pretty interesting and wasn't quite finished but it was 99% of the way there. it would've worked as-is and with 1 minor modification been perfectly usable. He used JK flipflops that toggle each time it is written to, so you could change between SSR1/2 and SSR5/6 but there was no way to know which set was in use! I have a complete schematic of the board now. As for the sound quality, it indeed is pretty poor on that SHAQ chip (it's actually SHAQ and not SHAO. Q = ceramic DIP, N = plastic I think) and I think there is a reason it does not have the "mozer" name on it. It is monotone and indeed sounds like it was made out of phonemes and I too thought it sounded very sp0256 like. Speaking of, I have a big collection of SP0256 and SP0264 chips and their associated speech ROMs, and have dumped all of those as well. hit up my email and I can send you an EPROM with several digitalker sets in it if programming them is difficult for you, so long as you're in the US. you can find it on my web page. search for my youtube username and it should come right up then the contact page. I have a lot of speech synthesis related stuff up.