How do I improve how my mixes sound on streaming services such as Spotify, SoundCloud and YouTube?
By Rob Stewart - JustMastering.com - Last updated on April 13, 2019
- While mp3 and other streaming audio technology has limitations, you can improve how your mixes sound over these formats.
- Loudness normalization is now used widely, giving more reason to create mixes with wide open dynamics which will improve how they sound when streamed.
I often receive questions about how to improve sound quality on streaming music and video sites such as Spotify, SoundCloud and YouTube. This article addresses those questions and my suggestions apply to any streaming service or any lossy file or streaming format (Mp3, Ogg, AAC etc.).
Some things to note about music streaming services:
- they are all different (some sites simply have higher quality sound than others)
- they use lossy encoding formats for streaming out of necessity, and the trick is how to work within the limits of the lossy format
- some use loudness normalization, meaning if your mix is too loud, the service will reduce its gain to align to a target
- your recording quality and mix quality will have the biggest impact on the resulting sound quality after encoding
- that last point is worth repeating - your recording and mix quality have the most significant impact on the resulting sound quality
Is there such a thing as Mastering for SoundCloud or YouTube?
I personally do not set out to master a track specifically "for SoundCloud" or "for YouTube". I follow the recommended practices for Mastered for iTunes which - in my view - apply to any lossy format (such as avoiding clipping and distortion, and keeping peaks below -1 to -1.5dBFS).
Streaming formats used by SoundCloud, YouTube and other websites will be here for awhile, so it is important to understand how to make the best of these formats. The spin-off benefit to improving your sound on SoundCloud and YouTube, is that - in general - you will achieve better recordings and mixes, too!
Improve the sound of your mixes on SoundCloud and Spotify
This advice applies to any music streaming site. If you search the internet to learn about the sound quality of SoundCloud, you will find many forum discussions where countless people express their frustration at how their mixes sound after they post them to a streaming service, such as SoundCloud. The issue is not SoundCloud's or Spotify's fault. It is possible to achieve a relatively high sound quality on SoundCloud and other streaming audio sites.
At the time of this writing, SoundCloud converts all uploads to a lossy format for streaming purposes. While lossy formats have limitations, they are a necessary evil because of today's internet bandwidth limitations. Despite those limitations, it is possible to achieve a relatively high sound quality when streaming your music.
First, it is important to understand a little bit about how lossy compression works. Using Mp3 as an example - and without getting too deep into details - Mp3 uses something called "perceptual coding" to compress audio. At a very high level, all that means is that the encoder analyzes your music and removes pieces of it that it doesn't think you can hear (elements that are hidden by more prominent elements in the mix). For example, at 128kbps, the encoder will remove anything above 16kHz (many people cannot hear too far beyond that point). It may also remove parts of your mix that are masked by stronger elements. To get a sense for pieces that get removed, please check my "Issues With Lossy Formats" article.
Optimize your mixes for any streaming service
Perceptual coding schemes typically respond poorly to complexity such as distortion because it can often be harmonically rich, and wide-bandwidth, and that is the root of most of my suggestions. Having said that, here are my recommendations for achieving the very best sound quality when posting your mixes to any streaming service, including SoundCloud:
- Create the cleanest recordings possible, prior to mixing (avoid unintentional distortion, significant background noise etc). Noise, distortion, rumble and other non-musical information in your recording will create more work for the encoder. Strive for the purest, clearest recordings. You can always add color selectively to certain tracks but avoid capturing dense, congested color or "mojo" on every single track. Remember that less is more, and the contrast of "clean" and "colored" tracks in your mix is what will make it interesting, while making the encoder's job a little easier.
- Follow my critical mix tips. Remember, the more you do to achieve a clear, lively, and dynamic mix, the more chance it will sound good when converted to a lossy format like Mp3. Keep your mix open, dynamic and clean.
- Avoid heavy use of saturation and distortion. Distortion is often wide-bandwidth in nature which presents several problems for perceptual coding (i.e. how does it know what parts of the distortion are intended to be "musical" and which are not and could be removed?).
- Avoid overly dense or congested mixes. Craft mixes in such a way that each of the elements are clearly defined (have their own space in the mix) and that there is no congestion. The most challenging scenario is a mix with several heavily distorted guitars. In that case, it is necessary to carve out enough sonic space for each guitar so that they can clearly be heard. If they sound like they are blending into one big guitar sound, then a lossy encoder may have challenges with it.
- Avoid heavy limiting and dynamic compression. I say this for a few reasons. First, it adds harmonics (distortion, which is wide-bandwidth), but it can sometimes fill in the sonic spaces that the encoder is looking for with those harmonics, forcing the encoder to make more compromises. Many mix engineers like to compress to "glue" a mix together, and that's certainly valid in some cases, however when it is overdone, there simply isn't enough space in the mix for the encoder to figure out what elements can be safely removed during encoding.
- Keep peaks below -1dBFS, I would even suggest -1.5dBFS or even -2dBFS if you are working from a 24-bit file. Since many encoding schemes do not handle peaks that are at (or near) 0dBFS very well, you can end up with a lot of distortion on the transients.
- Consider lowering the gain of your mix. Pull the gain on your mix down so that the average levels are below -12 to -16dBFS. Better yet, consider some of the newer "loudness monitoring" options out there such as metering that is based on Loudness Units Full Scale (LUFS), and set the "loudness" of your mix somewhere between -16 and -23LUFS (before mastering) and when self-mastering, aim for masters that fall between -12 and -16LUFS.
- Export your mixes in a lossless format such as *.WAV or *.AIF. If you export to a lossy format such as Mp3 from the start, then SoundCloud will be re-encoding from one lossy format to another (this is called "transcoding" and it degrades the quality even further).
- If (and only IF) none of the steps above have helped, consider low-passing your mix with a 6 or 12dB-per octave roll-off so that energy from 14-15kHz and above is reduced, and high-passing it at around 35-40Hz. In theory, this step should never be necessary but in the event that there is something unusual happening beyond the lowest/highest extremes, this pre-filtering stage may help improve things. Use caution and careful monitoring, though (there can be music way down at the bottom - the lowest A on a Piano is 27Hz!).
If you are skeptical about any of my suggestions, above, consider this fact: Mp3 predates the loudness war. It was developed in the early 1990s and was made available for widespread usage in 1995. The song "Tom's Diner" by Suzanne Vega is considered the "Mother of Mp3" because Karlheinz Brandenburg used this song (among several others of course) to fine tune his Mp3 enoding scheme. "Tom's Diner" was released in 1987 - well before the so-called "loudness wars" ever started - and also long before mainstream home recording became as popular and advanced as it is today (in 1987, many home recordists used cassette-based 4-track multi-track machines - how far we've come!!). Mp3 has surely evolved also, but it is important to note mixes with very heavy dynamic compression, saturation and "maximization" - that are often heard today - were more of a rarity in 1987.
How does Loudness Normalization impact how music sounds on streaming services?
Several streaming services now employ loudness normalization. For example Apple Music uses "Sound Check" which adjusts all songs to a set target gain level of -16LUFS and other services have targets of their own. Loudness normalization eliminates the perceived "benefit" that high gain will have on your master because if the gain is higher than the streaming service's target, it will only get reduced downward to meet that target. Therefore, any processing you use to maximize the gain (excessive brightening of the tone, heavy limiting, "maximizers" etc.) will only harm your mix (and your music!)for no real advantage. This means that now more than ever, mixes with wide open dynamics will sound clearer and more musical than heavily compressed and limited mixes.