Quoted from slochar:There are components that affect the speech timing on the S&T board - I had 2 EBD's side by side and one was just slower by a bit and it drove me crazy because it wasn't how I 'remembered' EBD being - that's the one that got sold, and the other one happily is pitched 'better' to my ears. Possibly the differences are down to something like this.
Now, what those components ARE is a good question. If it's the actual TMS5200, you might be out of luck unless you can get a bunch of those to swap. Someone at some point should figure out how to mod the sample data to work with the more readily obtained TMS5220.
Back when I actually had a tube a TMS5200(kick myself for not buying all I could when I could) every NOS chip with the same lot/date codes sounded different when tested on the same board. Most where slight differences but some where darth vader, much slower/lower pitch and others where more towards the alvin in the chipmunks faster / higher pitch. Maybe the leg rot that happens on these chips gets up inside of the chip package and effects something or just they just always sounded different from day of production... not sure
A TMS5220 almost works dropped in but most phrases are garbled. Its like it is pin compatible but the instructions are different. Having the speech ROMs modified to use the TMS5220 would be awesome as you can still get that chip relatively easily. I don't even know if that is possible to do tho. Not having a TMS5200 because the original's legs rotted away is a reason many of these boards are dead. I have a box of them without speech chips.
Atari used I think both the TMS5220 and TMS5200 in the video game Gauntlet. They probably had the right development tools tho that might be just lost to time.