What game are you trying to work on?
Here's the rundown:
- Most main dialogue and stuff that uses 12x12 or 16x16 font, is almost always SJIS. The reason for this, is the huge majority of the system card bios (rom) is actually just a sjis table of 12x12 and 16x16 characters. The bios has a get char function that takes a sjis value and size, as the argument.
- most CD 2.0 games use no compression. You can see the sjis plain as day in the data tracks (usually only 1 data track is the script and code, the last being a redundant data track, and any in between are either adpcm tracks or graphics for cinemas. The reason they break them up and interleave them, is to cut down on track seeking for segmented cinema loads).
- most CD 3.0 games and later gen games DO employ compression. All of the compression schemes that I've seen, are LZSS based. Usually nothing fancy. But you can't see a lot of the sjis text because of this compression.
- sjis is a two byte encoding. Very few games have single byte ascii support, at least for main text routines. Japanese sjis two byte still takes up less than English 1byte encoding. So assuming the game only supports 2byte encoding, and you don't know how to do ASM hacking (which for CD games is a pain in the ass, unless you find a way to get more ram - like upgrading the 'CD' to a higher level card), you won't be able to fit the english translation back in without cutting it down dramatically. The font spacing will also look like the infamous 't e r r a a n i g m a' translation.
- ASM hacking for CD games (especially 3.0 games) is pretty advanced. It's not uncommon for CD games to treat different parts/load of the game as completely different game engines. I.e You have to make multiple asm hacks. And it's possible free resources that you exploited to put in the hooks and new asm code, are in different areas and/or different in size for different areas of the game (Ys IV translation had this problem. So does Spriggan Mark 2).
- There isn't always room for a replacement font, let alone for new code (we ran out of space for the Super Raiden hack and that didn't even have replacement font routines).
What would be nice, is if there was a new system card with a few more ram space (even just 8k of ram) - for making translations. You could cheat and use the SuperGrafx for extra ram (24k extra ram), but how many people have SGX let alone SGX+CD. Myself and a handful of other people in the world.
Not sure what you goal is, but CD hacking on the PCE is not a beginners task. There are exceptions, but you'll have to look for those exceptions.