You paid for the lease, you recorded the verses, you dragged the vocal on top of the beat, and it sounds like two different songs playing at once. The beat is huge. The vocal is sitting on it like a sticker. Every tutorial you find is about mixing stems, and you do not have stems. You have one finished beat and a handful of vocal takes.
That is not a smaller version of the stems problem. It is a different problem, and most of the advice out there gets it wrong because it treats the beat like a raw instrument.
Why the vocal sounds pasted on
The beat you bought is a finished record. The producer mixed it, balanced the 808 against the kick, tucked the hi-hats, then mastered it: limited, loud, and filling the whole spectrum from 30 Hz to 18 kHz. Nothing in that file left room for a voice, because the producer did not have your voice when the beat was made.
Your vocal is the opposite. It is a single dynamic mic in a bedroom, peaking somewhere around -12 dB, with room reflections and breath and a couple of plosives. One file is a polished product, the other is raw material. Put them next to each other and your ear hears the seam immediately. That seam is what people mean when they say a vocal "does not sit".
The two usual fixes make it worse. Turning the vocal up until it wins just makes the beat sound small and the vocal loud and dry. Slapping EQ on the beat to "make space" undoes decisions the producer made on purpose, and compressing a file that has already been through a limiter gives you pumping, not glue.
Rule one: leave the beat alone
The beat's EQ, dynamics and tone are the producer's work and they are done. Level and placement are the only things you should change on it. Everything else happens on the vocal and in the relationship between the two.
This is exactly how Bandmixr treats it. When you upload one stereo beat next to one or more vocal files, the engine labels the beat as Instrumental / Beat on its own and switches it to passthrough: no high pass, no EQ, no compression, no reverb, no repairs. Only its level and a protective output ceiling apply. The vocals get the full chain. If the automatic label is wrong, say you uploaded a bass stem and a drum loop rather than a finished beat, change it by hand and the engine goes back to mixing them as instruments.
Get the files right before you touch a fader
Most vocal-over-beat mixes are lost before mixing starts, in the files themselves.
The beat should be a WAV, untagged, at the sample rate the producer worked in. If your lease only came with an MP3, know what that means: the master you make from it can never be cleaner than the MP3 you started with, and pushing an MP3 through a loud master exposes the codec's artifacts in the cymbals. WAV, FLAC or MP3 explains why. If the WAV costs 20 dollars more, that is the cheapest upgrade in this whole process.
The vocal should be dry. No reverb, no delay, no compression printed into the file. Pitch correction is the exception: if you tune, print the tuned take, because tuning is part of the performance and it is not something a mix should second-guess. Record at a level where the loudest chorus line peaks around -12 dB, with the mic close and the room as dead as a duvet and a wardrobe can make it. How to Record Vocals in a Rehearsal Room applies word for word to a bedroom.
Every file starts at zero. Do not trim the silence at the front of the vocal, and do not export the vocal with the beat mixed in "for reference". A vocal file that contains the beat cannot be mixed, for the same reason a stereo bounce cannot: Can You Mix a Single Stereo File?
What actually gets done to the vocal
Since the beat is fixed, all the mixing lives in the vocal chain. Here is what a rap or sung lead over a finished beat needs. The numbers are sensible starting points if you do this by hand; a good engine, human or otherwise, adapts them to the voice and the beat in front of it.
High pass around 80 to 100 Hz, adapted to the voice. The beat already owns everything below that, and low rumble from the mic stand only muddies the 808.
Resonance cuts where the room rings. Small rooms leave one or two narrow peaks in the low mids that read as boxy. Cutting them is what makes a bedroom vocal sound like a booth vocal.
Compression with a fast but not instant attack. Rap delivery is dense with consonants, and an attack of 3 ms clamps every one of them. Around 5 ms attack, 80 ms release and a 5:1 ratio keeps the words intact and stops the vocal from disappearing between lines. Shorter releases pump audibly against a beat that never moves.
A small presence lift somewhere between 3 and 5 kHz. That is where intelligibility lives, and it is where a rap vocal needs to poke through a beat full of hi-hats.
De-essing around 7 kHz, a bit harder than for a band vocal, because the presence lift above brings the S sounds up with it.
A very short room instead of a hall. Hip hop vocals sit dry and upfront. A tiny room reverb with a decay under a second glues the voice to the beat without pushing it back. A sung pop vocal can take a medium hall; a rap vocal cannot.
Then the level. Bandmixr balances the vocal stack against the beat as a group rather than file by file, so the beat keeps its punch and the lead stays intelligible however many takes you upload. That balance is the whole question, and it is easier to judge at matched volume than by feel. If you want the lead 1 dB hotter, the fine-tune faders are there for exactly that.
Doubles, ad-libs and hooks
Most rap sessions arrive as a lead plus five to ten extra files: doubles on the end of bars, ad-libs, a stacked hook. Label the main take as Rap vocal or Lead vocal and put doubles, ad-libs and hook stacks on Backing vocal. That distinction matters more than it looks.
Ten vocal files of the same voice at "normal" vocal level add up to a wall far louder than one voice. That is the single most common reason a vocal-over-beat project comes back with the beat sounding quiet and dull: the stack buried it. Bandmixr treats the backing takes as one stack, keeps them from piling up, spreads them across the stereo field, and keeps the lead in the middle. If your phone recorder saved mono takes as stereo files with identical channels, the engine spots that and still spreads them instead of stacking everything dead center.
Mastering a song whose beat was already mastered
The beat arrived at -8 LUFS or louder. Now there is a vocal on top and the whole thing needs a master. Two things to know.
First, do not reach for the loudest preset. The beat has already been through one limiter. A second, harder one on top adds distortion and Spotify turns the result down to around -14 LUFS anyway. Streaming at -14 or Standard at -11 keeps the punch the producer built. Pick Your Master Loudness walks through the three presets.
Second, leave headroom in the mix. If the beat was leveled so it peaks at 0 dBFS and the vocal is added on top, the mix bus clips before mastering gets a chance. Bandmixr's ceiling on the beat prevents that, but if you mix by hand, pull the beat down by 6 dB first and let mastering bring the level back. How Much Headroom to Leave Before Mastering has the details.
The short version
Leave the beat alone, get a WAV, record the vocal dry and close, label the lead and the stack differently, master at a sane loudness. That is the entire recipe, and every step of it is about respecting that one of your two files is finished and the other is not.
If you want to hear it done, upload the beat and your vocal files to bandmixr.com, pick Hip Hop or Pop, and compare the first mix against your rough at the same volume. The first download is free, so the test costs you ten minutes and nothing else.