Freesound Forums

I as a soundmaker think that there's something Freesound could do to help it's users better fulfill the world's sound needs.

If users could see for example top 100 most searched keywords in the Freesound search engine...

We'd know better what is nit enough of yet. And maybe do that instead of randomly recording stuff almost no one needs

Cheers!

unfa wrote:
If users could see for example top 100 most searched keywords in the Freesound search engine...

Stats on "searched for but not found" would also be useful.

In general on the open web, it is often depressing to find out what the top search terms are, as often they are related to the physiological (first) step on Maslow's hierarchy of needs.

I'd be interested to know what all registered users' top search queries are, because I suspect that anonymous search queries might be tainted by bots.

My guess of the top 10 is:
police siren
church bell
explosion
rain
thunder
clock
door
applause
drum roll
cat

@Timbre - good point!

ungoogleable? unfreesoundabled?

I have recently joined Freesound project as part of my master thesis in the Music Technology Group @ UPF. My work is very related to this thread. My idea is to study the searcher's behaviors and intentions. We are planning to connect the users of samples to the creators/uploaders of samples better through the analysis of query logs. Advertising the sounds for which no sound was available is one way to do this. I'm very glad that there are Freesounders suggesting this already!

(If you guys have any other suggestion regarding searching , ranking, search results quality, etc., I would love to hear your ideas)

My guess is that horror and vocal sounds would rank fairly high as well.

divaldo11 wrote:
(If you guys have any other suggestion regarding searching , ranking, search results quality, etc., I would love to hear your ideas)

Probably unrelated: I would like that it was possible to suggest tags to sounds, sometimes I find for example a gem of an explosion sound, that can only be found using "Trinitrotoluene" because the author didn't consider that most people would want to find it when searching for "TNT" or "explosion" or "explosive" (fictional example).

The scenario "people search for 'explosion' and there are many 'explosion' sounds" doesn't yet mean that any of them are good. The ranking of the sounds could be taken in consideration for this.

Another interesting question: how do users get more specific during their searches. Example:
1st search: "explosion"
2nd search: "explosion war"
3rd search" "gun shot"

These sequences could be isolated using time difference between searches or by using the action "user downloads sound" as a slicer (assuming this means that they were successful). It would be harder to estimate when a user fails at searching for a sound.

I agree - suggesting change of tags to be considered by uploader would be cool.

qubodup wrote:

Probably unrelated: I would like that it was possible to suggest tags to sounds, sometimes I find for example a gem of an explosion sound, that can only be found using "Trinitrotoluene" because the author didn't consider that most people would want to find it when searching for "TNT" or "explosion" or "explosive" (fictional example).

We have been discussing tag suggestion to the uploader for sometime. It think it's a good idea in involving the community and bringing together different perspectives and different ways of describing sounds. We are also implementing a synonyms table that contains polysemous words (ie TNT, Trinitrotoluene, explosion). Currently, it's aimed at the 'editorial staff' but we hope to be able to involve the whole freesound crowd to modify these synonyms in the future. Kind of like how Stackoverflow does it ( Tag Synonyms )

qubodup wrote:

The scenario "people search for 'explosion' and there are many 'explosion' sounds" doesn't yet mean that any of them are good. The ranking of the sounds could be taken in consideration for this.

This is probably one of the most important questions we are trying to answer. Ranking is a very subjective process. We want to know what constitutes a high quality result corresponding to a specific query. By quality, I don't mean just the recording quality. There are factors that plays into choosing a sound: # of downloads, description, format, ratings, etc. It would be nice to know the community's opinion on this.

i'm curious, would be interesting to see what people are searching for

qubodup wrote:
In general on the open web, it is often depressing to find out what the top search terms are, as often they are related to the physiological (first) step on Maslow's hierarchy of needs.

I'd be interested to know what all registered users' top search queries are, because I suspect that anonymous search queries might be tainted by bots.

My guess of the top 10 is:
police siren
church bell
explosion
rain
thunder
clock
door
applause
drum roll
cat

You missed "DJ Spin-maestro" of the list

Users which download my stuff won't be misguided `cos my tags are clear:

http://www.freesound.org/people/zagi2/sounds/179153/

Well tagged. I won't check that - I'm trying to keep my mind clean

Seeing some general info on searches and their subsequent results (was there a download?) could be handy. Slightly related (or unrelated)...

What I as an uploader am more interested in is how my sounds are actually used, and even how they are used. But I can't think of any way for the freesound mechanism to help with that. Ratings say nothing about what gets used. Download counts can also be very misleading, though they can certainly provide a gauge for the "minimum" number of people with a sound in hand.

There have been a few times I've done some vanity searching, using Google, to find how my sounds are being used out there in the wild, at least by those who bother to abide by the license. It works for me since I've used such a bizarre user name on freesound, so that I can search for my user name and freesound, and omit freesound as a site match, and get pretty good results. I figure I should do this every few months or so just to keep track of what new things are being done with my sounds.

What I think are some of the best sounds I've uploaded, that should be heavily used, may not get much notice. On the other hand, a few other odd-ball sounds get used over and over. There are some older recordings I made that are very clean, high SNR, that get such usage: a celery bite, a celery bite with subsequent chewing, and some up-close cricket sounds. Who'ld've thunk it?

But many of the sounds get used in games, and others in amateur videos posted to You-Tube. The previous discussion using "explosion" sounds as a fictional example made me think of this, because I do have some explosion sounds that I think are especially cool, and none of them are field recordings -- all are totally synthetic, created from scratch mathematically. Someone will pick one of my explosion sounds to use in a video or something, and I think it's probably not the best choice for their application, but oh well (to each his own).

It seems that everybody and their cousin is writing some game or another, or at least a mod for minecraft. And then there are the guys who make videos about games. And all of them want to use free sounds they can get on the web. A few actually take the time to offer attribution, and some even ask first.

I'm often surprised at what people use. There were some recordings I had of wood pieces being knocked about, and those have been used in various places. That one was a surprise to me. One person found that one of my vegetable sounds (cracking a carrot, or some such thing) made a perfect typewriter sound for some typing software, but I found that out from them offering a comment (not from online attribution).

For me, that is the sort of information I'm interested in, but I can't think of any good way for the freesound website to help with that, unless you want to take on the job of becoming a meta-crawler that is constantly scouring the web for uses of freesound materials, with or without attribution.

Thanks for this post. Pretty interesting elaboration.

1TBK1/2 regiert