Brought in from the thread on my idea for the version markers script and as a sound engineer very passionate about the Loudness topic, I totally support this idea.
Envato has already some of the ingredients of the recipe in place. Since it needs to analyze the incoming audio files to create the waveform displaying in the player – wherever that analysis is performed during the uploading process –, applying the LUFS normalization after that analysis is completed would be just a matter of adding one single line of code to the underlying engine putting together the material before exposing it to the public.
From a programming point of view, a FFMPEG command like the following one would be enough to programmatically normalize any audio file to the desired LUFS:
which manipulates input_file.mp3 to –13 LUFS, –0.3 dB True Peak and an 11 LU dynamic range and saves it as output_file.mp3.
Of course, through to the previous analysis that generated the waveform for the player, FFMPEG could also calculate the current LUFS of input_file.mp3, leaving unaltered all those files that have already a loudness lower than the target one.
The technical implementation would be very easy and, provided the fact that the waiting time for a track to be approved is already several days, this is definitely not a real-time process, so its implementation is far from being critical for the Envato’s server resources.
I have just run both the analysis and the normalization on a 4-minute track using a MacBook Pro:
– the analysis was completed in about 10 seconds
– the normalization was done in about 18 seconds
This would be a piece-of-cake of a task for Envato’s servers!
Thanks Luigi for accepting my invitation. Your ideas and knowledge are a great asset. It’s amazing to me, that all of that can be achieved with one line of code!
It looks easy simply because it is easy!
The magic is all done by FFMpeg, a command line tool that I use on a daily basis to analyze and process my audio and video files. It is possible to install it on a web server and either use it as a command line or through the numerous PHP libraries out there.
I have just installed it on my website – on a shared server host by SiteGround – and the normalization for the same 4-minute track was completed in 24 seconds.
I assume the whole Envato ecosystem runs in a pretty large server farm, and it would be interesting to see how marginal this type of processing is compared to the daily workload the server has to cope with.
@LuigiPulcini
Really glad you joined this thread, mate!
(thank you @criskcracker for inviting our advanced friend!)
Luigi, you are the one who definitely can help this idea come to life!
Envato Team should take a look at your technical proposal. It looks like there is not much complexity into your implementation.
Spotify is using ReplayGain. Do you think it is any better than the method you described?
It looks like ReplayGain’s approach is adjusting the level of an audio file that is played back, based on factors that are embedded into the file as metadata after the original audio file is analyzed.
So, basically, the difference between the two approaches, is that FFMpeg would create a new audio file that was re-mastered to the target Loudness, while ReplayGain allows any player compatible with that technology to adjust in realtime the level of the audio file without touching the actual file content.
In my opinion, it would not be a bad idea to re-master the preview files because that would make them independent from the player you are using: the audio file would always have the same loudness in any player you are using.
I don’t think this is the best way to do it because video editors who are importing preview files into their projects and doing all the editing before putting the final would notice a significant volume difference when they bought the final files wouldn’t they?
IMO keep the downloaded previews the same as submitted by the author, just adjust the playback in realtime on the site (as we’ll all be using the same player on the site right?)
annnnnd… while we’re at it, can we implement some sort of auto-watermark system like the other guys do? Making your own is kinda a pain in the butt. The upload process at another popular site takes about a quarter of the time as compared to here.
You have a point here, @RobertReid.
I did not consider that filmmakers are using preview files to edit. So, yes, being crucial to maintain the same loudness between the preview and the downloadable files, the only option left is incorporating in the Tsunami Player some sort of ReplayGain technology.
I will study a little more what it actually does and get back to you.
For the audio watermark, manipulating the file with FFMpeg would be pretty easy.
I don’t see why it shouldn’t be possible to implement ffmpeg in a way that the manipulated file would be played by the aj player, while the original mp3 would be downloaded when clicking the downloaded button… Wouldn’t that be just like two additional lines of code?
Yes, but at the same time that would also mean duplicating the whole preview archive. In terms of server space, the ReplayGain seems to be more cost effective.
Hello all. Thanks for your comments and ideas here. We regularly reach out to customers about their experience using our platform and in particular over the last few months have been interviewing and surveying customers in the audio space to understand their needs.
To date this is not something that they have prioritized. They have talked about discovery (how they find tracks), organization and the visual experience of trying to find the right fit.
We love that you are thinking about things like LUFS and standardizing the experience for customers but at this time it doesn’t appear to be a high priority for customers. That said, this is something worth thinking about and so we’ll make sure that this is brought forward to the teams working on AJ/Elements as something to consider.
I don’t think this is something customers could be expected to know about or want. But I’m sure that if some kind normalization of track levels where implemented customer would get a better experience shopping from Audiojungle. This is one of those things the customers don’t know that they want.
Hello @jamesgiroux!
Thank you so much for replying in this thread!
We all wanted to get an official response regarding this topic. So here it is!
I can totally understand that surveys do not indicate that customers are asking about LUFS normalization.
However, I have to say that we don’t really need such data to realize that this feature is extremely important, and we shouldn’t wait until the last moment to implement it. Because this is already a world standard, no overstatement here
All the leaders in the music industry are doing normalization and they are not going to stop it.
Spotify, Youtube, Apple Music, and many others.
I have no information whether or not they did any surveys on their sites. But we can see that what they are doing is good for customers. And so for creators. I think we can learn from these music giants.
I would agree with @Hyperprod here. Customers are not sound guys, they shouldn’t know anything about LUFS. They will simply get a much better experience if we do that. Isn’t it something we want? I know Envato want it! So all we!
Thank you very much for saying this, James!
“we’ll make sure that this is brought forward to the teams” - this is a great promise. And this is what we all wanted here!
I and many other authors really hope that this idea will be in a sight of development teams as soon as possible! Teams probably know what this is all about, but it would be cool if they could also take a look at this thread as there are many great points on why this idea is beneficial for buyers, authors and the market overall.
Thank you again, James, for giving us an official response and we hope to see some amazing changes or news!
Guys, did anyone pay attention (or am I the most observant )? In the general search, the track now sounds quieter in volume (I noticed this yesterday). It would be interesting to verify if it corresponds to the normalization of YouTube.
)? In the general search, the track now sounds quieter in volume (I noticed this yesterday). It would be interesting to verify if it corresponds to the normalization of YouTube.