Soundcloud is in trouble.
Three years ago, in August of 2017, Soundcloud’s investors approved an emergency bailout package of $170 million and fired their CEO.
Despite this, and when combined with the mounting issues of the platform, I’m still rooting for them.
They provide independent artists, or even just a kid with a laptop, a platform to share their music, network and build an audience.
However, my support for the platform ends where shady music production practices begin.
In July 2020, Soundcloud launched their new mastering service, costing $4.99 per track. Members on the Pro Unlimited plan receive three free masters each month, with a charge of $3.99 for more.
Sounds great, right?
Only I have recreated this new service in 5 minutes, using free software that is 15 years old.
But, let’s begin with a brief overview of the elusive craft of mastering.
What is mastering?
Mastering is the dark art of audio production. It is the last stage of a song's production cycle, consisting of two steps. The first is ensuring a track is sitting at the correct volume level for streaming. The second is balancing a song to provide accurate playback across a variety of mediums.
When I saw Soundcloud's mastering service using purposely vague terms, like “advanced technology” and “powered by Dolby”, it aroused my curiosity.
These phrases aim to provide a level of confidence by signalling that mastering is too complicated to understand. But, I've noticed in music production how this marketing method mystifies the process. This lack of transparency is the thread I began pulling on.
Here are my technical gripes for the geeks of the world. Please skip to the demonstration chapter if you're not a proud audio nerd like myself.
Soundcloud streams music in 128 Kbps MP3 files. 128 Kbps refers to the compression used to reduce the size of an audio file. The highest quality MP3 file is 320 Kbps, with WAV format being superior. If you were to compare the same song in MP3 to a WAV file, even untrained ears would hear the difference. This is because when compressing a song to MP3 format, reducing the file size comes at the expense of stripping much of the frequency information. When presenting a sound processing service, showcasing the results as low-quality MP3 files is questionable.
You could argue that both files are the same extension, making for a fair comparison. 128 Kbps is Soundcloud's default streaming format, so why not showcase a real-world scenario? This seems reasonable; however, allow me to share my two other considerations.
Gain matching refers to matching the volume levels of two audio files. A tenant in mastering is increasing the volume of a song to the appropriate playback level. In theory, by making something louder, you're mastering. The louder of the two files will always sound better because we perceive a higher volume level as being superior. The correct way of demonstrating a before-and-after mastering example is to ensure both tracks are near the same average loudness level. That way, you can hear the differences aside from volume alone. Changing the loudness should never be the judge of good mastering, nor should a service exploit the loudness as a trick to emphasise its capabilities.
If mastering is making changes to a track to ensure it sounds accurate across a variety of mediums, shouldn't those adjustments be tailor-made? Absolutely. This is why humans still occupy the mastering engineers role. But, as I'm about to show, Soundcloud's mastering is the same settings, copied and pasted over each of the examples. Their marketing copy leads us to believe the service is a custom 3-piece suit for our songs, but unfortunately, it's an off-the-rack solution. And you know what they say about off-the-racks? They never quite fit.
I have an issue with my internet browser. Firefox tends to download any audio I stream online. Oops. I should contact them about this. But, as I have both the pre- and post-master files in my downloads folder, let's put my theory to the test.
Here's what you're about to see and hear. There are six audio channels that each contain the before and after mastering files from Soundcloud. The channels titled 'Mine' are my mastered versions of their pre-master files. The two plugins at the bottom of the screen are applying the “advanced technology.” The channels located under each of 'Mine' titled 'Soundcloud's' are Soundcloud's mastered versions. I switch between Mine and Soundcloud’s in quick succession as this is the optimal way of hearing sonic differences. For the best comparison, wear good headphones and close your eyes or look away from the screen so you can't see when I change.
Can you hear a difference? I can. Mine contains a little more high-frequency content, which is most noticeable on the last of the three tracks. I spent 5 minutes replicating their “advanced technology,” and I believe these results are adequate. This experiment began for my Discord server, but I thought an article would be a more appropriate format. I have verified my findings many times to ensure they're factual and I'm not spreading misinformation.
My mastering chain is easy to replicate for anyone who knows their way around a D.A.W. Stating that it took 5 minutes isn't a testament to my skill, but to Soundcloud's generic new service.
Conclusion of the Demonstration
I have emulated Soundcloud's mastering service with “advanced technology” using 15-year-old, free plugins. Download the Ableton rack (free) and save your $4.99. If you don't use Ableton, copy this image over to your plugins of choice.
For anyone interested in the art and craft of mastering, I have a free 1-hour tutorial on YouTube.
P.S. For anyone who wants a better mastering chain, download this Ableton rack. Admittedly, this option isn't much better than Soundcloud's offering, but at least it shows you the makeup of a basic mastering chain and provides a good jumping-off point for experimentation.
I'd like to play devil's advocate and address a counter-argument to one of my above points. The example tracks all have a similar RMS level. RMS stands for root, mean, squared and is the average loudness level of a piece of music. All three of the examples sit between -20 and -24 RMS. I find this curious as Soundcloud has selected three tracks of similar levels. You may have noticed that each song sounds comparable, all electronic, featuring the same instruments and sounds. My mastering chain may work, as it is Soundcloud's preset option for all electronic songs with a -20 RMS level. If this is true, it doesn't reduce the validity of my above points, as I presume you need to tell Soundcloud which genre your track fits within. From there, the service can use the given preset to shape your work. The original average loudness of the song would not matter either, as Soundcloud brings everything up to an RMS of -10 to -14dB.
The equalisation curvature of each track contains striking similarities. The Equalisation curve is the balance of frequencies across a song. When combined with the above theory, it leads me to believe a strategic selection has taken place.
The choice of the three pieces of Electronic music is an interesting one. Soundcloud is for all genres of music, artists and songs. Why the exclusion? Why showcase your new technology on three interchangeable tracks? Why not display a diverse selection of music that would emphasise the service as being for everybody? I'm guessing this music works best for the demonstration, as to why though, I cannot say.
My central critique is a lack of transparency. The discrepancy between what Soundcloud both states and demonstrates when compared to my own conclusion is why I wrote this article.
I like that premium users are getting more as part of their subscription. Even though I dislike preset mastering, I'm not against providing a starting point for those new to the subject. Having mastering built into the streaming platform is convenient. I'd imagine many artists wouldn't care about the service so long as their music sounds better as a result. My only hope is that this doesn't lead to those newcomers believing Soundcloud's service as being legitimate mastering. I fear this could result in a reduction of work for the real engineers, as those beginners could argue: Why should I pay for mastering when Soundcloud does it for free?
The algorithms may be Dolby's intellectual property, thus making it illegal to share them. In which case, I would like Soundcloud to acknowledge the preset nature of the service and not propose tracks receive individualised treatment.
What do you think? Are there any flaws in my demonstration or statements? I have gone over everything with a fine-tooth comb, but every day, I realise how much I have left to learn.