Hi everyone,

2025 has arrived and this will be a very special year for Freesound! We will let you know soon about a number of things happening this year, but before that, it is time for the 2024 in numbers blog post. We will show some general statistics similar to those shown in previous years’ posts. But to make it a little bit different, this year we have also included a section with statistics related to search queries. So let’s get started: the number of new sounds uploaded during 2024 has been of…

56,964 new sounds!

which corresponds to…

1157 hours of audio!

There have been ~7,924 more uploaded sounds during 2024 when compared to 2023. However, in terms of hours of audio hours, the increase has been of 23 hours only. In 2023 there was an unusually high average sound duration, but in 2024 we see an average sound duration of 73 seconds, very similar to 2022 and other previous years. Therefore the numbers are not surprising.

Here is the Creative Commons license distribution of the newly uploaded sounds:

In 2024, the percentage of Attribution NonCommercial sounds has remained similar with previous years, but there’s a significant change in Attribution sounds. Attribution has increased significantly, surpassing Creative Commons 0 for the first time (45% vs. 44%). Historically, Creative Commons 0 accounted for about 60% of contributions, while Attribution was around 25%. It’s too early to determine the cause of this shift, but one possibility is that sound contributors are responding to concerns about generative AI by becoming less willing to share sounds in the public domain. What are your thoughts? Share them in the comments!

With the new additions from 2024, Freesound now currently hosts a stunning amazing total of 670,954 sounds. The total audio duration is of 476 days and 4 hours. Here is the evolution of the total number of sounds since the beginning of Freesound, and the prediction for the future:

We will still need to wait a couple of years until we reach 1M sounds, but we will surely surpass the 700k mark during 2025 🙂

Here is a tag cloud of the tags of the sounds uploaded during 2024:

If we compare this tag cloud to last year’s one, we’ll see that there are number of big words that have appeared which were not common (e.g. soundtrack, industrial, underground, dark, loopable). The usual suspects are still relevant (e.g. field-recording, ambient, drum, synth), but their relative size has been diminished. We have already seen many times how, even if many sounds are uploaded every year by many contributors, it is definitely possible for a single user uploading many sounds to change the distribution of the most used tags. And this is what is happening this year. User looplicator has contributed an immense quantity of sounds this year (see below), and many of them use some of the tags that I listed above. Here you can see a list of the packs uploaded by looplicator.

Tired of reading? Do you simply want to listen to some sounds? Here are some of the better-rated/most-downloaded sounds of 2024:

And after this intermission, let’s continue with the classic chart of the users who have contributed the most sounds in 2024:

	Username	# uploaded sounds		Username	uploaded time (hours)
#1	looplicator	13183	#1	looplicator	133
#2	Hewn.Marrow	1740	#2	Philip_Goddard	74
#3	Sadiquecat	1337	#3	klankbeeld	63
#4	klankbeeld	1276	#4	csengeri	45
#5	Tian_Yueyao	793	#5	Sadiquecat	33
#6	josefpres	616	#6	GregorQuendel	26
#7	ShangusBurger	585	#7	EtherAudio	22
#8	FOSSarts	486	#8	kevp888	17
#9	GregorQuendel	483	#9	josefpres	16
#10	jerry.berumen	479	#10	euangile	15
#11	BlondPanda	427	#11	twiciasty	12
#12	AudioCoffee	377	#12	tim.kahn	12
#13	hardcore scm	377	#13	eudeline.paul	11
#14	Mastersoundboy2005	333	#14	felix.blume	10
#15	Artninja	317	#15	Hewn.Marrow	10
#16	memeshift	314	#16	jerry.berumen	9
#17	Ragnar59	312	#17	bruno.auzet	9
#18	James_KuKu	279	#18	LolaMoore	8
#19	kevp888	275	#19	gadesound	7
#20	LolaMoore	272	#20	makeitsound4me	7

As we already hinted before, this year we have a clear “winner” in both number of sounds and number of audio hours. But hey, this is not really about winning or losing, thanks everyone (not only those appearing in the table) for all the contributions!

And what about downloads? The number of sound downloads (including packs) during 2024 was…

18,786,432 downloads!

Even though this number is slightly lower to that of last year, this is still an incredible amount of downloads. All in all, users have downloaded more than 254M sounds and packs from Freesound!

Now some extra general statistics: In 2024, 14k messages were sent, 1k forum posts, were written, 444k sound ratings were made, 26k sound comments were written. Sound comments have decreased by 40%, but sound ratings have been increased almost a 100% (so they have almost doubled!). Sound ratings were already at an increasing tendency, so this is great news that the tendency is confirmed. One possible explanation is that the new UI released at the end of 2023, invites users to provide more ratings (which was one of the design goals!). We’ll have to see how these numbers continue evolving in the following years.

The term cloud below shows the most common query terms that have been used when searching in Freesound during 2024:

The top 10 terms are very similar to those of previous years, with slight variations in the ordering: wind, rain, explosion, music, whoosh, footsteps, woosh, click. However, there are some new terms vivaldi, spring, and bird which were not normally at the top. We believe this is due to some bot activity as almost all of these queries originate from a single place, so we should not read this as the Freesound community suddenly growing an interest for Vivaldi. The term-cloud above is generated using raw search data, but a deeper analysis on this data can reveal some things interesting for the community. This is something that we usually don’t have time to look into, but this year we have computed some extra nice statistics about how users search in Freesound, so we’ll finish this blog post with a special section about that :).

Let’s start by looking at the question of when do people search? Search activity on Freesound is a round-the-clock phenomenon, though some patterns emerge when we look at the data. Unsurprisingly, activity peaks during weekday afternoons and early evenings (UTC), when the number of queries per hour reaches close to 15,000. Conversely, the quietest times fall during the late-night hours, dropping to about 7,000 queries per hour.

To break this down further, the platform processes an average of 4 requests per second during peak hours and around 2 requests per second overnight. While the overall distribution aligns with global waking hours, the consistency of activity at all hours highlights Freesound’s international user base and its role as a creative tool across time zones.

We also looked at what type filters are being used in search queries. When it comes to narrowing down search results, tags take center stage. Tags accounted for over 57% of filter usage, far ahead of other filters like licensing (15.4%) or file type (5.9%). This likely reflects the importance of keywords in categorising and locating specific sounds in Freesound’s extensive library.

As can be seen from the pie chart, filters related to Creative Commons licenses are also widely used, with license:”Creative Commons 0″ being a particularly common choice. This underscores the importance of freely reusable content for the Freesound community.

Interestingly, filters such as bit depth (16 or 24 bits) and sample rate (44.1 kHz or 48 kHz) are also frequently used. These technical attributes are crucial for users working in professional contexts, like sound design or film production, where specific audio quality standards must be met.

And what about the sorting of search results? The vast majority of search queries use the default sorting behaviour, Automatic by relevance, with over 81% of requests. Other sorting options, like Date added (newest first) – (11.9%) -, Downloads (most first) – (2.1%) -, and Ratings (highest first) – (1.7%) -, were far less common, though they reveal specific user priorities. For instance, sorting by newest entries suggests a focus on exploring fresh content, while downloads likely indicate interest in popular sounds. What is somehow surprising is that the option of sorting by sound average rating does not hold a bigger percentage, as this is a useful tool to get high-quality sounds! Maybe it means that the average sound quality in Freesound is high enough so that there is no need to sort by ratings.

To deepen our understanding of what users are looking for, we draw from a recent paper presented at the DCASE 2024 Workshop (an academic event that we usually attend, mainly dedicated to computational methods for the detection and classification of acoustic scenes and events). This research examined Freesound search queries to classify their content and identify patterns. A key finding was that users’ searches often fall into diverse, overlapping categories, visualized in the sunburst chart below:

• Music-related queries (e.g., instruments, genres, moods) and natural sounds (e.g., animals, water, fire) are consistently popular.
• Human sounds and sounds of things (e.g., mechanical noises, alarms, explosions) highlight Freesound’s value for media production projects.
• The source-ambiguous sounds category, including thuds, beeps, and whooshes, stands out as an example of Freesound’s appeal for crafted and designed audio. These sounds often play a pivotal role in sound design, where their flexibility makes them invaluable.

The study also revealed unique aspects of search vocabulary. Users frequently employ specialized terms like jargon (e.g., “riser,” “stinger”), abbreviations (e.g., “bgm,” “atmo”), and descriptions of intended use (e.g., “error,” “alert,” “game over”). This illustrates the specificity of Freesound’s audience, many of whom are professionals or hobbyists with clear creative goals.

And finally, let’s look at the query length. Interestingly, most queries on Freesound are short, typically only a few words. This matches findings in broader search behavior studies but raises intriguing questions. For instance, would it be beneficial for Freesound to adjust its default search settings to encourage users to submit longer, more descriptive queries? Could such a change unlock richer creative possibilities for the platform’s users? Or does the current “few words + filters” system just work and there’s no need for other strategies? Or something in between? These are all very interesting ideas, which essentially address the question of How do we humans talk about sound? How do we describe it and how would we like to interact with it? Well, surely we won’t find quick answers to these questions, but we can say that Freesound is definitely a great place to seek for them.

Aaaaaaand that is all for this year’s post, thanks for reading and we hope you enjoy a 2025 full of sounds!

frederic, on behalf of the Freesound Team
the special section discussing user search queries is a contribution by Benno Weck, thanks!