What value of compression would be typical for the spoken word?

[cfg]

What would be a good general starting point for compression on a recording of the spoken word or singing at speech level?

 

[LazerBeakShiek]

Bricked. Brick wall.

I always start with the effect(compressor) dimed. I wanna see what it can do first. Then blend it in.

 

[Farview]

4:1 medium attack and release. Set the threshold so it does what you want.

 

[bouldersoundguy]

Compressing spoken word is like compressing anything else. You do it for two reasons. First is to make it sound good. You generally want a nice consistent level, some variation but not too much, so it's comfortable to listen to. Second is to fit it into the technical parameters of the delivery medium, which is probably at least partly meant to do the same thing.

 

[bouldersoundguy]

Looks like I misread the OP. Try 4:1 with moderate attack and release. You might try automating levels upstream of the compression.