FarmerPotato

Right, there was one coding table in the S&S chip, TMC0285. It changed significantly for the 5100 and 5200.

Off topic for this thread, but yeah, scanning a page in grayscale at 600dpi TIFF will make each pdf page 33 MB. 400 dpi about 15 MB. Black-and-white scanning mode fails if the text is faded, esp from dot matrix printer. So I scan in grayscale, then run a Threshold filter to get acceptable B&W text out. That turns unreadable text into crisp text, permits compressed TIFF, and gets it to ~250K. Still 400dpi. Finally OCR, which has been working fairly.

For the expansion connector, the unpopulated solder points are still there on my units. Schematic shows it's mostly piggybacking on the installed PHROM. (Phrase ROM.) I'd like to have a socket so I could look at any PHROMs I find.

Are you interested enough to design a PCB? 5220C should work ok.

Ha. I see in the database spec a field for "Number of bits unused in last byte". A major part of the speech editor's work was to clean up the empty parts. For instance, zeroing out the volume between words. (I read several docs that explain this.) I see that these tools were constantly under development. The Fortran/Assembly code here for "CSEDIT" has 1981 change logs. (CS=Compressed Speech) Reference to "FREDIT" for Frame Edit, based on CSEDIT. A version of the CS or FR editor in Forth exists in the Texas Instruments Records at SMU (got photographs). I haven't found any code for the analysis of recorded speech to LPC. But there is a high-level description of the analysis--in a programmer's manual for the TIPC speech card. All this stuff was in storage since 1986. I doubt there is any DSR source preserved. The frame editor commands are in Fortran and Pascal. Assembly code in CSEDIT is the low-level routines for coding and uncompressing LPC, then speaking it. Comment says "SVC CALL TO THE SPEECH DSR".

I looked into the University of Arkansas connection. They did use the 990 barcode software for research thru 1987. So it was delivered. The BOM included two 5220A chips to upgrade the 5200 on the board. I read the professors names, because Cliff Easthom (my dad's co-worker on Magic Wand Reader) put them in a sample book file saying "hello, Dr Alan Vanbiervleit." And "Hello, Phil Parette". The outcome was published as Instructional Uses of Barcode Technology (Apr 1987) which says that they produced barcodes to help the disabled learn first aid and cooking recipes. Also related: Proceedings of the South Central Technology Access Confererence (Dec 89). Lists TIGER Communications TIGERbook in Appendix B. This was a book of 1000 phrases for daily living. They begged Texas Instruments for assistance in filling out hundreds of phrases. I want to know if TIGER was helped by the U Arkansas folks. ----- To introduce Barcode Factory 3, Cliff's sample book had just a few phrases, turned into allophones. The bar codes and text were output with a page layout , also defined in the book file. The output was probably a troff file for typesetting. (Incidentally, my dad collected manuals for a lot of word processors and formatters. More on that later.) The Magic Wand had a ROM with LPC for 127 allophones, plus some LPC phrases and sounds.* I guess that there was a software routine to speak them on the 990 card. Still, the quality check (QC) was to download assembly source or object code to a 99/4A. The 4A then controlled the Magic Wand through a joystick+cassette cable. (MMC plug on the MW.) The database aka the "Speech Disk" file on the 990 stored up to 4 remarks from QC "Voters". I don't know what became of the 4A to Magic Wand cable or schematic. The 4A software for it disappeared a long time ago. I remembered some 4A floppies went into a locked cabinet. Wasn't in there when I got it open. *The Text-to-Speech software for the 4A was adapted from the 990 program, created by Kathleen M. Goudie, the team's linguist.

No. It would have to be written fresh. But it's all CRU bit interface. Couple of 259 registers. I can guess what it would look like--write a record, 5200 interrupts when it's ready for more. User can query status.

But now, Romans have bidets. The public toilets cost a euro. In Ancient Rome, you had to pay for the sponge on a stick as well. Maybe the boy attendant was on the first step of the hydraulic corps ladder. Future aqueduct inspector.

This is a schematic for a speech synthesizer card for the 990/10. It is dated 3-11-80 (March). I found it among my dad's papers. He worked for Texas Instruments on speech products from 1978-1984. Notably, managing the Touch and Tell, VOCAID, and Magic Wand Speaking Reader. The 990 Speech Synthesizer required a DSR to be installed (SYSGEN). It probably would have replaced a 980 in the speech lab. That 980 was used to synthesize speech before the Speak and Spell's 0285 was in silicon. At least two of these 990 boards were shipped outside of Texas Instruments. One was installed by my father on a 990 at Texas Tech University, along with the analysis/edit/synthesis software, and a recording studio set up like TI's. It enabled TTU to offer a course on speech synthesis. (I saw the TI Lubbock West Building recording studio only once). The second was for the University of Arkansas. There's a long story there, which I'm only just learning the full extent of. Schematics scanned at 1200dpi, grayscale. Logic Diagram 990 Speech Synthesizer TMS5200.pdf

I've had a few inkjet printers and experienced more trouble than anything. But this HP Color LaserJet 4600dn is still chugging after nearly 20 years. Yes, the color is bust after I used a cheap refill kit and got magenta spilled everywhere. Ok, so the fuser is replaced with a refurb cuz a plastic gear broke. But I originally got it refurb, at the price of its HP brand ink cartridges. (Free printer with your deluxe toner?)

I found a 2-page document from TI that recommends using the UCSD PASCAL package to develop assembly language programs for the TI-99/4. Must pre-date the 4A and the Editor/Assembler. Does anyone remember seeing it before? Posted in this thread:

I found this curiosity, a 2-pager on 'TMS-9900 ASSEMBLY LANGUAGE WITH TI EXTENDED BASIC'. Did it come with the Extended BASIC cartridge? Or maybe the p-Code peripheral? It might have been internal to Texas Instruments. It says TI-99/4 not 4A. Assembly Language With TI Extended BASIC.txtAssembly Language With TI Extended BASIC.pdf TMS-9900 ASSEMBLY LANGUAGE WITH TI EXTENDED BASIC TI Extended BASIC has the capability, when used with the Memory Expansion accessory, to load and execute TMS-9900 Assembly Language subprograms. Assembly Language routines may be entered and assembled using the TI-99/4 UCSD PASCAL(tm) Development System. This document provides an overview of the capabilities, how they are accessed from TI Extended BASIC, and how assembly routines may be developed. Assembly Language routines called from Extended BASIC may be used to make algorithms execute more rapidly, or to provide complex control of the video screen or the sound chip. Assembly routines are normally loaded in a relocatable form into a dedicated 8K byte block of the Memory Expansion. Using relocatable code allows routines to be written and used independently of the actual loaded address in memory. Also, combinations of several routines may be used from one BASIC program without concern for the load address. Assembly routines are called from the BASIC program by name. The actual address of the routine is resolved during execution. One or more entry names for a routine are defined when the routine is written. The 8K byte reserved block of the Memory Expansion may be used by Assembly Language. In addition, a portion of the other 24K bytes in the Memory Expansion is available depending on the size of the BASIC program being run. An assembly routine may or may not return control to BASIC. It may actually take control of the TI-99/4 and act as an entire application. In this instance, the entire memory resources of the TI-99/4 are available for use by the assembly subprogram. A set of utility routines is provided for Assembly Language subprograms to easily access unique hardware resources such as the Video Display Processor, and to pick up the values of arguments or return values to the BASIC program. A set of subroutines may be linked with an assembly routine to provide access to peripherals. With these routines, an assembly subprogram may easily execute I/O to a diskette, RS-232 or other peripheral. Three statements in the Extended BASIC language (CALL INIT, CALL LOAD, and CALL LINK) are provided to support the Assembly Language capability. The BASIC subprogram CALL INIT is used to intitialize the RAM Expansion for assembly subprograms. This subprogram ensures that the Memory Expansion has been attached. It then loads a set of utilities from ROM memory in the Extended BASIC command TMS-9900 Assembly Language Page 2 with TI Extended BASIC module. into the Memory Expansion. CALL INIT also causes any pre- viously loaded assembly routines to be undefined. Any subsequently loaded routines will remain defined until another CALL INIT is executed or the Memory Expansion is turned off. The BASIC subprogram CALL LOAD is used to load an Assembly object file into the Memory Expansion. Alternatively, direct data may be specified in the form of addresses and data in the CALL LOAD statement. An object file may contain relocatable or absolute address code. Direct data may only be absolute. Because of the difficulty in defining the entry point name from direct data, it is not recommended for general use in an entire program. However, if this is necessary, we can provide technical assistance to help accomplish this. Examples of CALL LOAD statements follow: 1. CALL LOAD("CS1") 2. CALL LOAD("DSK1.MYOBJECT", "DSK1.OBJECT2") 3. CALL LOAD(12000,04,01,02,0,255,255) The first two are examples of loading from a file while the third uses direct data. The BASIC subprogram CALL LINK is used to link to an Assembly Language subprogram. This statement specifies the routine to be called and any arguments to that routine. For example CALL LINK("NSORT",A(,)) could be written to sort array elements in row A(2,n) so that elements in row A(1l,n) are in ascending order. In this example, NSORT is the name of the routine. One argument is passed as two dimensional array A. Values in this array are changed and returned to the BASIC program. Development of Assembly Language programs to run under Extended BASIC may be done on the PASCAL Development System. This system includes a powerful text editor, macroassembler, and link editor. A set of macros is provided during assembly to facilitate accessing specific TI-99/4 hardware features such as the Video Display Processor, or the sound chip. An I/0 utility is available which may be linked with an assembly routine to provide access to peripherals such as disk drive, RS-232, etc. Although the disk format of the PASCAL system is different from that used by Extended BASIC, a utility is provided to convert a PASCAL object file to the format required by Extended BASIC. **UCSD PASCAL is a trademark of the Regents of the University of California.

In disassembly of some TI code , I found this: BL @SUBR CLR R7 JEQ L2 ... I was confused, but I looked up CLR and saw it does not change the status register. Apparently R7 is used by SUBR, but this code wants R7 to be 0 before L2.

Second that. I built ought on a wire wrap perf board once I had the Geneve2020 CPU module ready to plug in. Memory, 9901, 9902 and sundry went in wire wrap first. Im tempted to go back, or a combination of PCB with next steps in WW. So much trouble soldering my latest boards. Along the way, I learned some history of wire wrap. It is a superior technology! Connections, physically, are superior. Breadboard doesn't compare. . The standard Radio Shack WW hand tool is still reliable. (1980s). Got the parts to a Jonard electric tool, but still the hand tool is second nature.

I bought from him a keyboard compatible with my TI-928 terminal. It was in amazing shape and so an amazing value.