What is a reference track?
A reference track is a finished, released song that you load into your session and switch to while you mix. It gives your ears something fixed to measure against: if your bass is heavier than the bass on a record you love, or your vocal sits further back, you hear it straight away when you flip between them.
You aren't trying to copy the song. A reference answers questions you can't answer from inside your own mix, such as whether the low end is too big or the snare loud enough. Engineers use references while mixing and again while mastering.
Why use reference tracks?
Your ears adjust to whatever you're listening to. After twenty minutes on one mix, a top end that's too dull starts to sound normal, and a bass that's too loud starts to sound like the way the song goes. Play a well-mixed record for ten seconds and your ears reset.
Your room and speakers are the other reason. Small rooms tend to pile up bass at some frequencies and lose it at others, and cheap headphones exaggerate or hide parts of the spectrum. A reference plays through the same room and speakers as your mix, so the faults of your setup affect both, and the difference you hear between them is mostly the difference between the mixes.
Then there's the listener. Spotify, Apple Music and YouTube all normalise loudness, so your song will be heard at about the same level as the songs around it in a playlist. A reference at matched loudness is a close copy of how a listener will compare your track with everything else they play.
Reading the lab
The lab plays two things in time with each other. Your mix is built live from three loops (drums, bass and chords) with a fader each and two shelves on the mix bus. The reference is a full mix of loops from the same packs, pushed through a limiter like a release. Upload a song you like with the Your sound button and it becomes the reference instead. The file stays in your browser, and the lab loops its first 30 seconds, so trim the song to its chorus before you upload it.
The big screen is the spectrum of each track, averaged over the whole loop. Your mix is pink, the reference is blue, and each point is the energy in a third of an octave, so pink noise would draw a flat line. The shaded area between them is the difference.
Under it, the balance screen splits the sound into lows below 120 Hz, mids from 120 Hz to 4 kHz and highs above 4 kHz. For each band it shows how far your mix sits above or below the reference once both are at the same loudness, and the shaded strip marks 1.5 dB either side. The panel on the right explains what you're hearing, using numbers measured from the two tracks.
Match loudness turns the reference down to your mix's integrated loudness. A/B, or B on your keyboard, switches between them without losing your place in the loop, and the Listen to buttons put both through the same filter so you can compare one band at a time.
How to choose a reference track
Pick songs that you'd be happy for your mix to sit next to in a playlist. The closer they are to your track in style and arrangement, the more useful they are. A sparse acoustic song can't tell you much about a dense electronic track, because the two are meant to sound different.
Choose something you know well and have heard on lots of systems, in the car and on your phone as well as at home. You already know how it's meant to sound, so when it sounds odd on your speakers, you've learned something about your speakers.
Use two or three references and listen for what they share. If all of them have about the same amount of low end against the vocal, that's probably right for the style. If they disagree, you have some freedom.
Get the best file you can. A WAV or FLAC from a download store is ideal, and a 320 kbps MP3 is close enough for balance. Avoid recording audio from a browser or ripping a video, since those versions have been encoded at a lower quality and turned down by the site's own normalisation.
Then choose the section. Choruses are usually louder and fuller than verses, so compare your chorus with the reference's chorus, and loop eight or sixteen bars of each.
How to level-match a reference track
Do this before any other comparison. A released song has been mastered, usually with a limiter that made it several decibels louder than any mix in progress, and a louder sound seems better, fuller and brighter. Compare the two at their real levels and the reference wins for reasons that have nothing to do with the mix.
Loudness changes tone as well. Your ears are less sensitive to low frequencies at quiet levels, so when a track is turned up its bass seems to grow faster than its mids. A louder reference sounds better, and it also misleads you about how much bass it has.
The fix is to measure both in LUFS, the unit streaming services use, and turn the reference down by the difference. Say your mix reads −17.1 LUFS integrated and the reference reads −10.9 LUFS. The reference is 6.2 LU louder, so you lower its fader, or its clip gain, by 6.2 dB. That's exactly what step 3 in the lab does.
Measure the same part of each song, ideally the loudest section, over at least ten or twenty seconds. A free meter such as Youlean Loudness Meter shows integrated loudness, and Logic Pro comes with a Loudness Meter plug-in. If you have no meter at all, matching by ear is far better than not matching: flip between them and adjust until neither seems louder. The LUFS lesson explains what the meter is measuring.
Setting up a reference in your DAW
Put the reference on its own audio track and line it up so the section you want to compare starts at the same bar as yours. Then check where its signal goes. Most projects have plug-ins on the master channel, and if the reference passes through them it gets your bus compressor and limiter on top of its own mastering.
You can send the reference track straight to your audio interface's outputs, which in Ableton Live is the Ext. Out option in the track's Audio To menu. You can switch the master plug-ins off while you compare. Or you can put a reference plug-in last on the master channel, such as ADPTR Audio's Metric AB or Mastering The Mix's REFERENCE, which swaps the whole output for the reference at the click of a button and handles the level matching for you.
Whichever you choose, make switching fast. A shortcut that mutes one track and unmutes the other matters more than the method, because the comparison only works if you can flip in about a second.
What to listen for
Compare one thing at a time. Flip between the two, listen only to the low end, decide, and then move on. Trying to judge everything at once mostly tells you that the reference sounds more finished, which you already knew.
| What to check | Where to listen | What it tells you |
|---|---|---|
| Low end | Below about 120 Hz, kick and bass | Too much or too little weight, or the two clashing |
| Brightness | Above about 4 kHz, cymbals and vocal air | A dull or harsh top end |
| Vocal or lead level | Full range, against the drums | Whether the main part is loud enough to follow |
| Drums | Full range, the kick and snare hits | Punch, and whether the drums sit forward or back |
| Width | Stereo, then summed to mono | Too narrow or too wide, or parts that vanish in mono |
| Reverb | The gaps between phrases | Too wet, too dry or a tail that runs too long |
Listen one band at a time
The quickest way to hear a tonal difference is to filter both tracks the same way. Put a low-pass filter at around 120 Hz on both and you hear only the kick and bass; swap it for a high-pass at 4 kHz and you hear the cymbals, the top of the vocal and the bright edges of everything else. With the rest of the song gone, a 3 or 4 dB difference that was hard to spot in the full mix becomes obvious.
In your DAW, put the same EQ on a group that both tracks feed, or copy one EQ onto both. The high-pass and low-pass filters lesson covers the filters themselves, and frequency ranges explains what lives in each part of the spectrum.
Check the parts by ear
Some things a meter can't tell you. By step 7 in the lab the three bands agree within 1.5 dB, yet the drums are still 4 dB further back than in the reference. The bars compare each band with the mix's overall loudness, and the drums add to all three bands at once, so a 4 dB change moves each bar by less than a decibel. Your ears pick the difference out at once from the snare. A vocal works the same way: compare where it sits against the snare in both songs, and set its level by that.
Width and mono
Switch your monitoring to mono and flip between the two again. If the reference keeps its bass and vocal solid in mono and yours loses a synth or turns thin, you have a phase or width problem to fix. The mono compatibility lesson shows what to look for.
Comparing on a spectrum analyser
An analyser such as Voxengo SPAN, which is free, or the spectrum view in your DAW's EQ, draws the average balance of a song. Put one on your mix and one on the reference, or use an analyser that can show two signals at once, and look at the overall shape before the detail.
Most finished music slopes gently downwards from left to right on an analyser like the one in the lab. If your mix has a bump in the low mids that the reference doesn't, or its top end falls away much faster, that's worth a closer listen. iZotope's Tonal Balance Control compares your mix with a target range built from reference tracks, which is the same idea with less setup.
Use the analyser to find where to listen and your ears to decide. Two different songs never have the same spectrum, because they have different instruments in different keys, so a line that doesn't match isn't automatically a problem. A decibel or two across a wide range is normal; 4 dB or more in one region usually means something.
How to use reference tracks, step by step
- Choose two or three referencesPick songs in your style that you know well, and get lossless or high-quality files.
- Put them in your sessionGive each its own track, line up the matching section and route them so they skip the plug-ins on your master channel.
- Match the loudnessMeasure your mix and each reference in LUFS over the same section, and turn the references down by the difference.
- Listen to a reference firstPlay ten or twenty seconds of it before you touch your mix, so your ears are set to how the style should sound.
- Compare one thing at a timeStart with the low end and work through the table above. Flip quickly and listen to the same few bars each time.
- Fix the biggest difference, then compare againOne change often alters how the other differences sound, so recheck before moving on.
- Get back to mixingReturn to the reference every half hour or so, or when you can't decide something. Hours of constant switching wear your ears out.
References and the final loudness
While you mix, leave the reference turned down. Your mix will always be quieter than a released song, and trying to close that gap with a limiter on the master channel squashes the drums before the balance is finished. That job belongs to mastering, which the mastering chain lesson walks through.
When you do master, the reference helps you choose how hard to push. Streaming services turn loud tracks down to about −14 LUFS on Spotify and YouTube and about −16 LUFS on Apple Music, so a master pushed far louder than your reference won't play any louder for the listener. It will just have flatter drums. Streaming loudness has the details.
Common reference track mistakes
Comparing at different loudness
Turn Match loudness off in the lab and the reference beats your mix every time. That tempts you into a second mistake: turning your whole mix up into a limiter to compete.
Running the reference through your master chain
A reference that goes through your bus compressor and limiter is no longer the record you chose. Route it around them, or switch them off while you compare.
Copying the reference exactly
Your song has its own instruments and its own key. Aim for the same broad balance, the way the lab's strip allows 1.5 dB either side, and leave the small bumps on the analyser alone.
Comparing the wrong sections
A reference chorus against your verse will always sound bigger. Loop matching sections of both songs.
Using a low-quality file
A low-bitrate copy can lose some of the very top end, and if you match your mix to it you'll make your mix duller than the real record. Buy or download the best version you can.
Listening for too long
Flip for a few seconds at a time, then get back to work. After a while every comparison starts to sound the same.
Train your ears with the quiz
The quiz under the lab starts from a mix that matches the reference, moves one of the shelves by 5 dB and covers the screens. Both play at matched loudness, and you say whether your mix has more low end, less low end, more brightness or less. Eight rounds take a few minutes. If the highs are harder for you than the lows, listen to the bright top edge of the bass, which carries most of that band in this mix.
Questions people ask
How many reference tracks should I use?
Two or three. One reference can pull your mix towards its quirks, and with three you can tell what the songs have in common, which is usually what you're after. More than that and you spend the session listening to other people's music.
Should my mix be as loud as the reference?
No. A released track has been mastered, usually with a limiter, and a mix in progress hasn't. Turn the reference down to your mix's loudness to compare them, and leave the final loudness to mastering.
Can I use an MP3 or a streaming service as a reference?
A lossless file, such as a WAV or FLAC bought from a download store, is best, and a 320 kbps MP3 is close enough for balance. Avoid recording audio from YouTube or a browser: it has been encoded at a lower quality and turned down by the site's own normalisation.
Should I use reference tracks when mastering?
Yes. In mastering the reference helps you set the overall tone and judge how hard to push the limiter. Match the loudness there too, or the louder version wins every comparison.
Can I use a song from a different genre?
For one specific thing, yes. A pop vocal can be a good reference for vocal level even in a rock song. For overall balance, stick to music in your own style, because a hip-hop mix and a folk mix are meant to have very different amounts of low end.
Do I need a plug-in to use reference tracks?
No. A reference on its own track, a loudness meter and the mute buttons do the job. Reference plug-ins make switching faster and do the level matching for you, which helps if you compare often.