Currently, there are situations which tell you the sex of the speaker, such as dark nights and so on. Then there are other situations where that doesn't happen.
In any situation where the sdesc is obscured or when the sex isn't revealed due to garbing, you should get an indicator of the voice speaking to you.
Rather than a "faint shape says, in sirihish", it should be "a faint shape says, in masculine/feminine/androgynous sirhish".
When you can see the sdesc of the speaker, or, when they are, for instance, "a male in a mask", this wouldn't be nessessary.