If you’ve been on the internet or music production Tiktok lately, you may have noticed that there’s been an abundance of stem splitter tools. Stem splitters have made it easier than ever for producers to start remixing parts of their favorite songs without having to wait for an official sample release.
Music giants are taking notice-- Kanye, known for his obsession with samples, has announced a stem player in collaboration with Kano. It seems that stem splitting tools are set to become a production mainstay. Below, we’ll share everything you need to know about stem splitters so that you can get in on the action.
What Are Stem Splitters In Music Production?
Stem splitters are tools that analyze a piece of digital music and break it down into stems, or common audio groupings within a track. For example, you may have a vocal stem, drum stem, and instrumental stem. These stems can be played independently from each other and when played together, they make up the full sound of a song.
This is incredibly useful for anyone creating a remix of a track. Vocal stems can easily be manipulated to fit on top another beat, or used for sampling on their own. The possibilities are limitless with stems, and with the increased access from splitters, we’re likely to see more creative interpretations of our favorite songs.
Keep in mind that “stems” does not equate to “samples”. This is a common misconception and the words have been used interchangeably, but one is not the other. Stems are specific groups of samples. Splitters are designed to produce stems, but you won’t be able to further separate within a sample provided by one of the AI stem splitters listed below.
How Do Stem Splitting Tools Work?
Stem splitting tools essentially use a combination of time frequency masking and AI learning to separate tracks into their respective stems. The system is able to analyze different parts of a song based on frequency and group them together based on the pre-trained model of the source.
For instance, vocal parts are going to reside at a different frequency than the drums. Presumably, the stem splitter will then be able to determine which sounds go into which stems. Unfortunately, it can be tricky to determine which instruments are which, which is why AI training is part of the process.
Most of the stem splitters are fed model tracks split by a human to start understanding how frequency splits work within a particular sonic context. While AI stem splitters aren’t always on the dot while splitting the stems of today, there’s no reason to doubt their capability in the future given enough training time.
Services like Deezer’s Spleeter provide users with a 2, 4, or 5 stem separation model. They also allow users to train custom source separation models, if you’re into that sort of thing.
So, What’s Up With The Donda Stem Player?
A lot of the recent hubbub on splitters can be stemmed from Ye himself. The $200 Donda Stem Player is designed to “split any song into stems” controlling vocals, drums, bass, and samples. Take a look at the stem player in action here.
The bluetooth device provides real time looping and speed control, along with the ability to share mixes and customize colors with the tactile controls. Dubbed as the first tactile stem splitter, it will be interesting to see how this device affects the music scene.The Donda stem player is set to be shipped out along with the rest of West’s album, which dropped August 29th.
Top Stem Splitters: Which One You Should Use?
Are you ready to make a remix or flip some samples? Here are some of the top stem splitters you should consider to pull stems from your favorite tracks.
For the sake of comparison, I’ll be splitting the stems of “Save Your Tears” by The Weeknd and Ariana Grande so that you can hear the difference between the models. I’ll focus on extracting the isolated vocals, but keep in mind that some splitters may be better at processing certain sounds than others and vice versa. This general breakdown simply serves as a way to determine which splitter makes the most sense for your needs.
Splitter.ai
Splitter.ai utilizes Deezer’s aforementioned Spleeter to separate the stems of tracks. The service provides three stem-splitting models: 1) a 5 track stem split, 2) a de-reverb tool and 3) a two-track splitter that provides separation between a song’s vocal and instrumental. The 5 track split is completely free, and the other 2 options can be tested with a free trial, or yours for $9/ a month.
Here’s the vocal stem using Splitter:
LALA.ai
LALA.ai is designed based on the two-track model, solely separating the instrumental from the vocal track. The service allows you to select between mid, normal, and aggressive processing modes. You can use the service for free with up to 10 minutes of content per month. There are two paid packages at $10 and $20 per month with 90 and 300 minutes respectively.
I decided to set it to “normal” while running the “Save Your Tears” track through the service:
EZ Stems
EZ stems also utilizes Spleeter with the 2, 4, and 5-track model. You can choose between 5 audio outputs and also select an upper range EQ cutoff. That being said, your initial file size must be less than 5.2 mb, so you might have to do some compression before running a song through the splitter.
The free version generously allows up to 2 uploads each day. There are also 5 paid packages ranging from $5 to $100 a month. Note that you have to wait in a queue for your stems to be processed, so you might have to wait a bit longer while using this service.
Here’s the vocal stem of the track:
Clearly, stem splitting tools are here to stay. Thankfully, today’s technology has made it easier than ever to extract stems from your favorite tunes. Let us know how you’re using your newfound stems in the comments below!