Freesound Forums

We invite you to participate in an exciting research survey exploring the search behavior of users in digital sound collections. This study aims to understand how you would approach searching for sounds using natural language text queries, such as fitting a sound to a video, story, or image.

In this survey, you will be presented with different prompts designed to spark your imagination. Imagine you need to find the perfect sound for a specific situation, and write the search query you would use. There is no need to perform an actual search; we are interested in your thought process and the words you choose.

The survey will take approximately 12 minutes to complete. Your participation is invaluable and will contribute to a deeper understanding of user behavior in digital sound libraries. The survey is quick, engaging, and completely anonymous. We appreciate your time and insights!

Click here to take the survey or copy the following link: https://sound-search-survey.streamlit.app

This survey is conducted by the same team that is behind Freesound, Universitat Pompeu Fabra / Music Technology Group (UPF/MTG).

We also welcome any comments or questions you might have — feel free to share your thoughts below!

Hi cecav16339,

thank you for your comment. I've updated my original post to clarify that this survey is run by researchers at UPF/MTG, the same team behind Freesound.

Surveys like this help us gather essential data for research that would otherwise not be available. The information collected is completely anonymous, and we do not connect the answers with any personal information. Your participation is crucial in helping us understand and improve sound search tools.

Thank you for your time and insights!

Hi, I have participated in replying the survey.
One thing wasn't clear enough, between the first "search prompt", and the "wrong sound" resulting in a second prompt, and if I was expected to continue in a "natural language" with an acceptable reply/prompt of "oh that's not what I wanted", or if I had to redo a traditional new 'individual" prompt from scratch.

I mostly participated as if it were the first. How I would like an ideal search remembering the first prompt, and refining it afterwards.

Cheers!

For me all these questioning in English is too hard to understand what you mean with every question. to detailed questioning for nonnative speaker like me.
'
I have to use translate all the time and than I still do not understand what you mean.
I am really sorry

But

I would ask the Future Search Engine the following qwestion:

Geef me het geluid van de Maas bij Bokhoven op een zomerse avond. Graag geen boten.

Whitch means

Please give me the Maas river at the village Bokhoven in the Netherlands on a summer evening. Please without shipping ore boats ore there engine.

The result will be this;
https://freesound.org/search/?q=-shipping+-boat+Maas+bokhoven+summer+-ship&f=&s=Rating+%28highest+first%29&si_tags=0&si_name=0&si_description=0&si_packname=0&si_sound_id=0&si_username=0&d0=0&d1=*&ig=0&r=0&g=1&dp=0&cm=0&mm=0

I thing AI should analyse my question with include and exclude analyse. Like i normally do.

Finding the right sound is searching the sound you wish to find an than exclude wrong things stap by stap using eg -sun to exclude the sun…..

I hope you understand.

For me this is the fastest, accurate and most simple way to search.

In my opinion this is how to build every AI search system in general.

Regards

Hi @klankbeeld and @Sadiquecat,

many thanks for sharing you thoughts about the survey and the search experience in general.

I think since the survey uses an artificial setting and not a real search enginge, it is expected that some confusion might happen. We intentionally left some things open to not restrict anyone in their answers.

Thanks again for providing feedback!

Thank You for Your Participation!

We sincerely thank everyone who participated in our sound search survey. The survey is now closed. Your valuable insights and contributions are greatly appreciated and will help our research at UPF/MTG.

Thank you once again for your support. We hope to engage with you in future projects!