Freesound Forums

Hey guys, this is my first post on the forum.

So, I listened to today's random sound of the day, and like alot of other sounds I've heard on this site, it's very quiet and hard to hear without turning my monitors' volume up. I use normalization on all my uploads, for convenience, because if you don't want it to be that loud, then you can turn it down, and you can hear it clearly in the preview. Since these are sounds for use in other things, you're probably going to want a normalized sound anyway. If it's somehow possible to normalize sounds when they are uploaded, that'd be really useful, because so many sounds are soooo quiet. Also, incase the normalized sound is too loud, you could have a volume control. Anyway, that's just a thought I had. Sorry if this post seems incomplete or something.

Thanks!

I believe it's unethical to interfere and tamper with the community's uploads.

Moderating uploads for poor descriptions and illegal audio is as far as we should go.

I have sounds that I deliberately uploaded at a low dBFS level, because they were intended to be that way, having them normalised without my consent would destroy my intention:

http://www.freesound.org/people/Headphaze/sounds/234508/

Something more acceptable would be to normalise the transcoded preview, so that the original file is left unadulterated, which creates a more seamless auditioning experience when cycling through sound results, YouTube does something like this on playlists to level out the contrast in volume between music tracks. That is more reasonable than forcing processing on peoples sounds.

edit: The above statement is indented for impartiality, still disagree with doing things like that. I'm an advocate of diversity and an adversary of the loudness war.

Headphaze wrote:

Something more acceptable would be to normalise the transcoded preview, so that the original file is left unadulterated, which creates a more seamless auditioning experience when cycling through sound results...... That is more reasonable than forcing processing on peoples sounds.

Oh no, please don't...

klankbeeld wrote:

Oh no, please don't...

Ah, that wasn't me approving the idea. That was me being impartial.

I still disagree with the idea, and it will never happen.

I had to edit my post to make that clear.

Do a search for normalization on the forum. We've had this request before...
There are many reasons not to do it

- bram

Bram wrote:
Do a search for normalization on the forum. We've had this request before...
There are many reasons not to do it

- bram

Actually I remember that conversation now you mention it.

Here it is: http://www.freesound.org/forum/bug-reports-errors-and-feature-requests/33429/

Its was nice to reminisce on that, thank you.

Bram alluded to this in his posts, but I will say it here very clearly.
I do not think Freesound is allowed to modify the sounds. (unless we change the user agreements/terms)

Freesound is a 'user content website'. We are only allowed to host the user's content 'as is', not allowed to modify it. - and this includes permanent changes such as normalizing the sounds.
Sounds must remain at Freesound exactly as the user uploaded them.
We are entitled to accept or reject a sound, but if we accept then we should not modify it.

There is another discussion going in another forum about improving tags. Although related, these points are fundamentally different because of the licenses and upload terms.
If I recall correctly, when uploading, the users will have no claims in relation to any 'metadata' (descriptions, tags) and other data they upload to Freesound (such as personal profiles). On the sounds, the terms are those of the chosen license and Freesound commits to hosting the sounds "as uploaded".

So I think we are Ok in relation to tags, especially as we are talking about ADDING - the user original content would remain.
As for the sounds themselves, they should remain as the uploader intended.

Yeah, I totally see your point. I wasn't really saying to straight up change uploads, but rather make it a recommendation, or have an option for listeners to do it in the preview player.

There could be the possibility that in the analysis we write the maximum loudness somewhere in the HTML file and that the player can "negate" this volume somehow. I.e. some automatic gain control when the user hits play.

This would be quite experimental, but it COULD actually work!

Maybe something for a hack-day!

Actually Alien, those terms don't explicitly forbid Freesound to modify the preview. The file is already modified, because it's transcoded for the lower file size in preview quality (which is technically a modification).

Speaking for myself, I see absolutely NO virtue in normalization of uploaded files or preview copies of them. The recording level as submitted is, for better or worse, what the recordist intended. If it's too loud for a particular listener, that would be apparent from the waveform image, so the listener could take pre-emptive action before playing. True, it would be helpful to have a volume slider in the player facility for each file on FS, but that's something very different from a blanket normalization that is inflicted upon all submitted or displayed files.

A particular problem appears to be that a huge number of people are conditioned to maximum-level-normalized sounds through listening to masses of popular / synthesized music, in which it is the norm (sic) for such material to be 'normalized' and thus actually homogenized into quite an unmusical state. For good reason, classical music is rarely normalized in that way, and the dynamic range and the low level of quiet music is usually respected.

I myself record natural soundscapes (which require similar treatment to classical music), and where normalization is necessary I do that myself; that occurs only where there is a particularly big dynamic range, so that a max-normalized file is at the highest level it can go without clipping - but it would generally still be mostly quiet, and indeed, in some of my more recent recordings (not on FS), even after max-normalizing, their dynamic range is such that a playback volume lift of anything from 3dB to 12dB is required (6dB and upwards necessitating a warning from me about possible damage to equipment without sufficient power handling headroom)!

Indeed Philip, I also make my own volume setting and I have a reason for that. Please do not normalize my sounds in the pre-view. Thank you.

I'm not talking about normalizing sounds.

I'm talking about automatic, optional gain control for our players that works in a per-sound way.
I.e. if it's a silent sound, turn up the gain, if it's Really Loud, turn down.
And, optional.

It wouldn't touch the original sounds. Everything stays the same.

- bram

Hi Bram,

I do understand well. I see the pro and contra for this.

Take this for example; https://www.freesound.org/people/klankbeeld/sounds/271679/

It is a realy quiet situation. If someone is listening this in the gain contoled situation it could be out of it's context.

But I do understand what the pro is for that option in pre listening for other sounds. A lot of people complain about the volume of some sounds.

What ever it will bring, i'm still a happy user and uploader.

Philip Goddard wrote:
A particular problem appears to be that a huge number of people are conditioned to maximum-level-normalized sounds through listening to masses of popular / synthesized music, in which it is the norm (sic) for such material to be 'normalized' and thus actually homogenized into quite an unmusical state. For good reason, classical music is rarely normalized in that way, and the dynamic range and the low level of quiet music is usually respected.

I would tend to question the relevance of these types of assertion to a conversation that has to do with auditioning sounds. Freesound is only tangentially associated with music, and I find it highly dubious that contentious production methods in “popular or synthesized” music have anything to do with it.

The question of what constitutes an “unmusical state” (or not) is pure artistic subjectivity, as might be evidenced from the fact that (for example) the modern major scale was once considered an unhealthy incitement to vice only for hymns to start being written in it a few centuries later, or indeed for harmonies not consisting only of perfect intervals to be anything more than walls of noise. In any case, I don’t see the connection to the way Freesounds are auditioned.

This is not intended as a personal attack, as I very much agree with the rest of your post, but found it spoiled a little by the above.

jamesabdulrahman wrote:
and I find it highly dubious that contentious production methods in “popular or synthesized” music have anything to do with it.

With all due respect, I pretty sure Philip was referring more to the conditioning of certain users to expect loud audio in every aspect of their online experience, with little regard for the true dynamic nature of sounds in a diverse place such as Freesound.org

Agreed, Headphaze. Clearly it would have been better if I'd not elaborated my point, as we don't need side-tracks in this discussion!

Regardless of how it may be described, the point is that, as Klankbeeld says, automatically adjusting the volume of recording previews does not make sense, even though some people would want it. It's really up to individual recordists to set / adjust the level of their recordings so that a sensible normal volume setting reproduces the recordings at something like the intended level for the listener, or, where there is a very big dynamic range, to inform the listener how much higher than normal the volume needs setting for the particular recording (this would be necessary with certain of my recent recordings).

There is one way that auto volume adjustment could be brought in, in a way that shouldn't put anyone's nose out of joint - if it were introduced as an opt-in feature for uploaders themselves, who could then opt in to having that feature enabled for their recordings. That, then, would ensure that the feature would be applied only appropriately, and nature recordists like me and Klankbeeld, and of course plenty of others, could then be sure that their uploaded recordings would still be presented at something like their correct level.

thanks Philip for this very clear way of explanation and good suggetion.