Preferably JLPT1 (3-4 is useless..)

I heard from a Korean friend that her Korean book apparently has something like that. I searched the internet, but when I search vocabulary for listening, I just get listening, and vocabulary as separate things of course.

The thing is, when it comes to pictures, the vocabulary becomes completely different. It usually ends up having to do with positions, appearances and shapes. If you could get a list and study them all it could be a huge benefactor I believe.
For instance, here's an easy word that you should know if you took JLPT2 or maybe even 3 - まるがお
I didn't use kanji for a reason. In speech you can't see kanji so I think these words should be learned in hiragana.

Raise up the difficultly a bit, you get words like たてなが, ひしがた and so on.

Anyone know any lists that aim towards these types of visual words?

I'm not sure exactly what you're saying, but if you're looking for JLPT vocab (this guy only goes up to JLPT 2) written in kana, maybe http://www.spurrymoses.com/jlpt/2/index.html would help. There don't appear to be sound files, though.

Thanks, but I'm not looking for a JLPT vocab list, but a Listening's vocab list.

Basically, I want something that focuses on words you hear in 聴解, not the 語彙 in the first 漢字 section or the 読解.
See, in Listening, a very large amount of pictures are faces, stances, clothes, etc.
There's also stuff like graphs where you ought to know words like 傾向/上回る/徐々に/etc which also come up on listening (and also reading)

Hello!

I found a site you listen to vocabulary words along with English definition (while you read the list!)

http://www.tanos.co.uk/jlpt/skills/vocab/

Unfortunately, the guy speaking (at least JLPT4 files) has a horrible gaijin accent. Useless for listening to for JLPT native-speaker listening training.

The lists are good though they may be better formatted elsewhere.

I don't think what you want exists. AFAIK the vocab lists list vocab that appears in all sections of the test, not just what's tested in the 語彙 section. The only way to get what you want would be for someone to transcribe all of the audio from previous tests and then run it through an analyzer like mecab.

It would be a pretty useful list, but very labor intensive to make.