Repository files navigation

Quoted fromGeneral Instrument Application Note 0505D: CTS256A-AL2 Code-to-Speech Chipset

Description

The Code-to-Speech chip set consists of two chips: the SPO256A-AL2, anallophone-based single chip speech synthesizer, and the CTS256A-AL2, an8-bit microcomputer programmed with a letter-to-sound based algorithm.This chip set translates English characters into LPC synthesized speechsounds.

SPO256-AL2

The SPO256-AL2 is General Instruments's standard allophone chip and isbased on the SPO256A speech synthesizer. This synthesizer consists of a10 or 12 pole second-order cascaded LPC filter, a controller, and a16-bit ROM in which 59 allophones (speech sounds) and five pauses arestored.

CTS256A-AL2

The CRT256A-AL2 is a PIC7041 whose on-board ROM is masked with ourcode-to-speech algorithm. This algorithm converts English text (in theform of standard ASCII characters) into SPO256A-AL2 compatible allophoneaddresses, using letter-to-sound rules.

I honestly hadn't expected the level of interest that was shown in thisboard when I posted it onr/electronics.So I hadn't prepared a repository. This effort is therefore a littlerushed, and I beg your forgiveness for any obvious in retrospect errors.

The schematic is based largely on the schematics included in the "TechnicalData" sheets that used to be stapled to the back of the Radio Shackblister packs. They were also included in the Archer Semiconductor ReferenceGuides, specifically the1988 edition.

Those schematics appear to have been based in turn on those included inthe aforementioned GI Application Note.

Since I wasn't expecting anyone else to want to use this, I've customisedthe schematic to suit myself. There's no parallel port interface, and theserial interface is programmed not by onboard DIP switches, but by thecontents of the exception word EPROM.

The active filter was cribbed directly from David C. Alford's articlein theDecember 1985 issue of Micro Cornucopia, "Speech Generation With Radio Shack's SPO256".Although I seemed to have it working well enough when I breadboarded thecircuit, the first version on the PCB was pretty horrible. It sounded likeit was clipping - a lot. It turned out to be a dumb mistake on my part.Just because the LM324 is spec'd to work on a single supply, the circuitwasn't designed for that. Without a negative supply, I was losing theentire negative half of the signal. How on earth I didn't hear that onthe breadboarded version, I have no idea. I toyed with the idea ofpulling V- from the MAX232A and regulating it down to -5V; the currentrequirements of the LM324 were modest enough I could have gotten awaywith it. But I couldn't quickly find a through hole negative regulatorthat didn't draw much more current than the LM324 would have. I had afew ICL7660s in my parts box, so I went that route instead.

The board layout was inspired by theCTS256A-AL2 Text-To-Speech Board.It wasn't until I saw that one a couple of years back that I realized thatmy oh-so-much-larger solderless breadboard version could be tamed intosomething more reasonably sized.

In thegerbers subdirectory, you'll find all the Gerberfiles, as well as a zipped compilation suitable for submitting toJLCPCB, which is where I had my boards made.