Your Voice. Their Track. One Record.
Human mixing for singers using AI-generated instrumentals.
You generated the instrumental in Suno. You sang over it. What came back sounds like two things happening at once instead of one song, and you can hear it even if you can't name it.
That's not your voice. That's the mix.
Send me the instrumental and your vocal takes. I put your voice inside that track, master the result, and send back a finished record that's ready to release. Your actual performance, not an AI's impression of it.
Why this is a different job than mastering
A Suno export is already finished. It was mixed and mastered assuming nothing else was going into it. The midrange is occupied. The whole thing is compressed to sit at streaming loudness. There is no room left where a voice would normally live, because the model never expected one to show up.
Put a home-recorded vocal on top of that and it floats. It sounds pasted on, because that's exactly what it is.
The work is carving space in the instrumental where the voice needs to sit, matching the tone of a bedroom recording to a track that never saw a room, and rebalancing everything so the master still holds together afterward. Then mastering the result.
Mastering polishes a finished song. This is the step before that, the one where the song becomes finished in the first place.
What you get
A professionally mixed and mastered stereo WAV, loudness and true-peak verified to current streaming targets
Built by human ears in a real DAW session. There is no automated service that does this job, and there is no preset for a voice.
One revision included
Your Mix Dossier: a written breakdown of what I did to your vocal and why. Where it needed help, what the instrumental fought me on, and three specific things to do differently when you record the next one. You should finish reading it knowing more about your own recording than when you sent it.
Turnaround: 3 business days for Red Lab Library members, 5 business days otherwise.
Pricing
Includes one lead vocal plus up to two additional tracks. A harmony, a double, a backing part, whatever you've got.
Additional vocal tracks beyond three: $25 each Second song in the same order: $179
Red Lab Library owners: $10 off, always.
Not sure your recording is ready?
Vocal Readiness Check, $49.
Send me your files before you commit. I'll listen, check that everything lines up, and tell you straight whether it's workable or exactly what needs fixing first. If it's ready, the $49 comes off the price of your mix.
If you're confident, skip it and order the mix. Most people should.
Here's what happens if you skip it and I open your files and find they aren't workable: you get the same punch list. Fix the issues and resubmit at no extra charge. If you'd rather walk away, I refund everything except $49 for the review.
Either way, you never pay for a mix I can't deliver.
What I need from you
Read this before you record. Two minutes now saves a week later.
1. Sing to the exact instrumental you're going to send me.
Generate your track in Suno. Download it. Record to that file. Do not regenerate, re-export, extend, or make a new version afterward. Suno renders are not identical between generations, and a vocal recorded against a different version can't be aligned to the one you send. This is the one that ends orders.
2. One continuous take per part, start to finish.
Record the whole song in one pass. Don't punch in, don't stitch takes together, and don't send me the first verse from one file and the chorus from another.
You don't need to hit record at the exact instant the music starts. A few seconds early or late is fine, I'll line it up. What matters is that the take runs continuously against the track from wherever it starts, so the timing stays consistent all the way through.
Flubbed a line? Sing the song again rather than patching that one spot.
3. Send dry vocals. Then, if you have one, send a processed version too.
The file I work from has to be dry: no reverb, no delay, no autotune, no compression, no EQ. Whatever you print into the file I can't take back out, and it fights everything I need to do. Monitoring with effects while you sing is fine, just don't bake them in.
Here's the useful part. If you've already dialed in a version you like, with your reverb and your tuning and whatever else, send that alongside the dry one and label it clearly. Not so I can mix it, but so I can hear what you think your voice should sound like. Taste is the hardest thing to guess at, and one file showing me your intent saves us a revision round.
No idea how to do any of that? Skip it. Send the dry file and tell me a reference track. Works just as well.
4. WAV files, matching sample rates.
WAV only. No MP3, no M4A, no voice memos, nothing sent through a text message. 44.1 kHz or 48 kHz, and every file at the same rate as the instrumental.
5. One file per part.
Lead vocal in one file. Harmony in another. Double in another. Don't bounce them together and don't send a rough mix with everything already combined. Name them plainly: Lead.wav, Harmony.wav, Double.wav.
6. Headphones, and a quiet room.
Headphones, not speakers. If the instrumental is playing out loud while you sing, the mic captures it, and that bleed causes phase problems I can't fix. Wired earbuds are fine. Bluetooth is not, because of latency.
Room noise, the TV in the next room, the air conditioning, traffic outside: all of it survives into the finished master. You don't need a studio. You need a quiet room and something soft on the walls.
7. Tell me what you're going for.
One or two reference tracks, or a sentence about the sound you want. Dry and upfront, airy and washed, tight and modern, whoever's records you keep coming back to. Optional, but it makes the mix better and saves a revision.
How to send files
Dropbox, Google Drive, or WeTransfer. Email attachments compress audio and text messages destroy it. Send the link with your order or reply to your confirmation email.
Who's doing the mixing
I'm Josh Gilliland, founder of JG BeatsLab. Five of my own AI-assisted bands are actively releasing. I wrote Unlock Reaper, the book on mastering AI music. I review 30 to 40 tracks a week as a SubmitHub playlist curator, so I hear the difference between a finished record and an export every single day.
I also published a blind test where my own mastering was scored against two automated services and the raw Suno export, with the file identities stripped. I lost one of the four and published that too, because a test you only report when you win isn't a test.
When your mix comes back, it was my ears on it, start to finish.
The honest fine print
I don't pull vocals out of a finished Suno track to replace them. Stem separation on a mixed AI export leaves artifacts you can hear, and I won't charge you for a result I wouldn't put my name on.
I don't fix a performance. Light timing and pitch work is part of the job. Rebuilding a take that's substantially out of tune or out of time is a different service, and I'll tell you before I take your money.
I can't promise a competitive record from every source. Some Suno instrumentals are too dense, too limited, or too artifact-heavy to sit a real voice in cleanly. If yours is one of them, I'll tell you what I'm hearing rather than hand you something and hope.
No open-ended engagements. No subscriptions. You own everything, always.
Questions first? Email me: josh@jgbeatslab.com. Founder-run, so you're emailing me, not a support desk.
— Josh Founder, JG BeatsLab