Mastering Service Test: Loudness, Tonal Balance, and Translation

Mastering Service Test: Loudness, Tonal Balance, and Translation

The best mastering service test is not "which master is loudest." Send the same mix to each service, level-match the returned masters, then score loudness control, tonal balance, punch, stereo stability, and translation on real playback systems. Once the masters are matched in volume, the strongest service is usually the one that keeps the song clear, musical, and consistent instead of simply making it louder.

Loudness bias is the trap. A louder master almost always feels better during the first few seconds because the ear reads extra level as extra excitement. That does not mean the master is better. It may have less punch, harsher highs, a weaker low end, or less dynamic contrast after streaming normalization. A useful test removes the level advantage so you can hear the actual mastering decisions.

Why Loudness Tricks Work So Well

The human ear is easy to fool in quick A/B tests. If one version is even a little louder, it often feels clearer, wider, punchier, and more finished. That is why mastering comparisons are dangerous when you flip between files at their delivered level. You are not only hearing tone or depth. You are hearing volume bias.

Streaming makes the problem even more important. Spotify's artist support documentation explains that Spotify uses loudness normalization during playback and adjusts tracks toward -14 dB LUFS according to the ITU 1770 standard. Spotify also recommends keeping masters below -1 dB true peak for lossy formats, and below -2 dB true peak if the master is louder than -14 dB integrated LUFS. That means a very loud master may not play louder on Spotify. It may simply be turned down while keeping the damage from the extra limiting.

The test should answer a different question: after the volume advantage is removed, which master still feels best? That is the master you can trust.

Prepare One Fair Source File

The comparison starts before you send anything. Every service needs the same source file and the same instructions. If one engineer gets a 24-bit WAV with plenty of headroom and another gets an MP3 or a limited rough bounce, you are not comparing services. You are comparing source mistakes.

• Export one clean stereo WAV: use the session sample rate and 24-bit if available.
• Remove final limiting: leave the mastering service room to work.
• Leave headroom: peaks around -6 dBFS are fine, but the exact number matters less than avoiding clipping.
• Use one written brief: send the same release goal, genre, and reference to every service.
• Use one reference track: different references produce different target decisions.
• Ask for the same deliverable: streaming master, high-resolution master, or both.

Do not give one service extra notes after hearing their demo. That turns the test into a revision contest. For the first comparison, judge each service's first interpretation of the same mix and brief. Revisions matter later, but first-pass taste tells you a lot.

Set Up the Level-Matched Listening Session

Import each returned master into the same DAW session. Put the files on separate tracks, line them up sample-accurately or as close as possible, and label them A, B, C instead of by service name. If you can, have someone else rename the files so you do not know which is which during the first pass.

Then measure integrated loudness and true peak with a meter. Youlean Loudness Meter's documentation defines integrated loudness as the average loudness over the full track or selected segment, and true peak as the detected maximum peak level used to avoid clipping and distortion. Those two measurements are enough for a practical test.

1. Measure integrated LUFS for each master over the full song.
2. Find the quietest returned master.
3. Turn the louder masters down until all versions are within about 0.3 dB.
4. Bypass master-bus processing in your DAW.
5. Listen to 10-20 second sections, not the whole song at once.
6. Take notes before revealing which service made which master.

If you level-match by ear only, you can still be fooled. Use a meter first, then make tiny ear-based adjustments if the measured match still feels slightly off. The goal is not lab perfection. The goal is to remove the easy loudness advantage.

The Five Scores That Matter

Once the masters are level-matched, score each version on five practical axes. Do not use one vague "sounds better" score. Separate the jobs so you know why one version wins.

Give each score a 1-5 rating. A service that wins three of the five axes at matched loudness is usually the real winner. If one master is louder but loses tonal balance, punch, and translation after level matching, it is not the better master.

What Loudness Should Actually Prove

Loudness is not meaningless. It is just not the whole test. A master that is too quiet for the genre may feel unfinished next to similar releases. A master that is too loud may lose transient impact and streaming quality. The right question is whether the loudness supports the music.

For many modern releases, it is normal to see masters louder than -14 LUFS. That does not automatically make them wrong. Spotify's own guidance does not say every song must be mastered exactly to -14. It explains how normalization works and gives true-peak recommendations to reduce distortion risk. The service should understand the tradeoff: louder masters may be turned down, while over-limited masters keep their reduced dynamics.

Use loudness as a pass/fail signal:

• Does the master distort when turned down to match the others?
• Does the hook still lift, or did limiting flatten it?
• Does the kick lose weight compared with the mix?
• Does the vocal get sharper after limiting?
• Does the true peak leave enough room for streaming encodes?

A good master can be competitively loud. It should not depend on volume to hide weak tonal decisions.

Tonal Balance Test

Tonal balance is where mastering taste shows quickly. Listen to the low end first. The kick and bass should feel controlled and intentional, not simply louder. Then listen to the low mids around the vocal, snare body, guitars, keys, and synths. If the master adds warmth by raising a broad low-mid area, it may sound expensive on monitors but muddy in the car.

Next, listen to the vocal presence and top end. A mastering service can make a song seem more finished by adding air, but too much high shelf makes sibilance, cymbals, and distortion artifacts more obvious. A better master opens the top without making the vocal hurt.

Translation Test

Translation is the most important score because listeners do not hear your master in one perfect room. Test the level-matched masters on at least four playback systems: studio monitors or good headphones, earbuds, phone speaker, and car. Do not change the song order every time. Take notes on the same sections: first verse, first hook, final hook, and the quietest section.

On phone speakers, listen for vocal clarity and snare presence. On earbuds, listen for harsh top and stereo balance. In the car, listen for low-end control and vocal level. On monitors or headphones, listen for detail, depth, and whether the master feels like a finished record. The winner should not be perfect everywhere, but it should fail the least.

If the same master wins on monitors but loses badly on phone and car, be careful. That may mean the service made a beautiful studio master that does not fit your audience's listening habits. If your listeners are mostly on earbuds and car systems, translation should carry more weight than studio glamour.

Revision Test

After the first-pass comparison, send one small revision note to the top one or two services. This is not about making the master perfect. It is about learning how the service handles feedback. A good mastering service can respond to a specific note without breaking the rest of the song.

Use one clear note like:

• "Can you reduce the sharpness in the hook vocal without making the song darker?"
• "Can you keep the low end full but make the kick/bass relationship tighter in the car?"
• "Can you give the hook a little more lift while keeping true peak safe?"
• "Can you make the master slightly less aggressive while keeping competitive level?"

Then compare the revision to the first pass. If the revision fixes your note but creates two new problems, the service may be using broad template moves. If the revision improves the exact issue and keeps the rest intact, that is a strong sign.

Blind the Service Names Before You Score

The best mastering comparison is blind for at least the first serious pass. If you know which master came from the expensive service, which one came from the fast service, and which one came from the engineer you already like, your notes will drift toward the story you already have in your head. That is normal. It is also exactly why loudness is not the only bias to remove.

Rename the returned files before listening. Use simple labels like Master A, Master B, and Master C. If possible, ask someone else to do the renaming and keep the key until after your first scorecard is finished. If you are working alone, rename the files, take a short break, then start the test from a new listening session instead of immediately judging while you still remember the file order.

Blind listening does not have to be complicated. You are not trying to run a scientific study. You are trying to protect the decision from price bias, brand bias, and delivery-speed bias. A master from a lower-cost service can win if it translates better. A master from a premium engineer can lose if it is too bright, too flattened, or wrong for the song. The point of the scorecard is to let the song decide.

Do the blind pass in two rounds. In round one, take fast notes on first impressions: vocal level, low-end shape, harshness, punch, and whether the hook feels exciting. In round two, slow down and score each category. Do not reveal the service names until both rounds are done. If the same master wins both rounds at matched loudness, the result is usually reliable.

Test Single Quality and Release Consistency Separately

A single master and an album master are not judged the same way. A single can be optimized for maximum impact because it has to stand alone. An album, EP, or multi-song rollout needs consistency from track to track. If you test a mastering service with one song and later use that service for a five-song project, ask how they handle song-to-song level, tone, and spacing.

For a single, score the strongest hook, the intro, the drop, and the final chorus. The master should make the song feel finished without making the most intense section collapse. For an album or EP, score the transition between songs. The loudest track should not make the next song feel weak, and the softest track should not feel like a mistake. Spotify's album normalization behavior is one reason this matters: albums can be normalized together so intentional song-to-song dynamics are preserved during album playback.

If you are testing for a multi-song project, send two mixes to the top service before committing to the whole release. Pick one dense song and one sparse song. A service that can make the dense song loud and the sparse song emotional, without forcing both into the same brightness and limiter shape, is usually more useful than a service that only has one aggressive sound.

Questions to Ask After the Test

Once you have a winner, ask a few practical questions before paying for the full release. The answers will show whether the service has a process or only a checkout page. You do not need a long technical interview. You need to know how revisions, deliverables, and quality control work.

• What files are included? Ask whether you receive a streaming WAV, high-resolution WAV, instrumental master, clean master, or alternate versions if needed.
• How are revisions handled? A good service should define how many revisions are included and what counts as a mastering revision versus a mix change.
• Do you check true peak and encoded playback? This matters when the master will be distributed to lossy and lossless platforms.
• Do you preserve album dynamics? If you are mastering multiple songs, ask whether the songs are judged together instead of as isolated singles.
• What should I fix in the mix first? A useful mastering engineer will tell you when a mix problem is blocking the master instead of over-processing around it.

The best answer is not always the most technical answer. Clear, practical communication is the signal. If the service can explain the tradeoff between loudness, punch, and translation in plain language, revision notes are more likely to go smoothly.

When to Stop Comparing

Do not turn mastering selection into an endless bake-off. Three services is enough for most releases. Four or more can create decision fatigue, and decision fatigue often pushes people back toward the loudest option. If two masters are close after level matching, pick the service with better translation, clearer communication, and a better revision process.

Also remember that mastering cannot fix every mix problem. If every master sounds harsh, muddy, small, or unbalanced, the mix may not be ready. In that case, mixing services may be the better next step before paying for another master. If the mix is already strong and you want a controlled release finish, mastering services are the correct path. If your vocal chain is the real issue before the mix even reaches mastering, vocal presets can help get a cleaner source tone before the final stages.

Mastering Service Scorecard

Use this simple scorecard. Fill it out before you reveal which service is which.

If one service wins the first five categories, the decision is done. If the top two are close, use revision quality and communication as the tiebreaker. If all three are close, your mix is probably strong enough that the choice matters less than you think. Pick the service that is easiest to work with and move on.

FAQ

How many mastering services should I test?

Three is enough for most releases. One AI or budget option, one mid-tier human option, and one trusted service gives you useful contrast without creating decision fatigue. If you already know the tier you want, test three services inside that tier.

Do I need a paid loudness meter?

No. A free or built-in loudness meter is enough for a practical comparison as long as it can show integrated LUFS and true peak. The goal is to level-match the returned masters and check for obvious peak problems, not run a broadcast compliance lab.

Should I pick the master that is closest to -14 LUFS?

Not automatically. Spotify uses -14 dB LUFS for playback normalization guidance, but many commercial masters are louder. Judge whether the master keeps punch, tone, true-peak safety, and translation. Do not pick a master only because it hits one number.

What if every master sounds worse than my mix?

That usually means the mix is not ready, the brief was unclear, or the services are pushing the track too hard. Go back to the mix, provide clearer references, or choose a service that is willing to do a more restrained pass.

Can I compare stem mastering and stereo mastering in the same test?

You can, but understand the difference. Stem mastering gives the engineer more control and may fix balance issues that stereo mastering cannot. If the stem master wins, you may be paying for partial mix help, not only mastering.

What is the fastest way to avoid volume bias?

Import every returned master into one session, measure integrated LUFS, lower the louder files until they match the quietest file, then listen in short sections. Do not compare files at delivered level if one is obviously louder.