Audio Levels and Loudness for Online Video Explained
Learn how peaks, loudness and dynamic range differ, how to read your meters, and how to keep dialogue and music consistent across all your videos.
Objective
Read audio meters correctly, tell peak level from perceived loudness, and set a mix that sounds consistent from video to video on any playback device.
Before you start
- A finished or nearly finished mix with dialogue and music
- An editor with an audio meter
- Headphones for checking the mix
Overview
Viewers rarely praise good audio levels. They notice bad ones. A video that is far quieter than the last one makes people reach for the volume knob. A video with a loud music bed makes them reach for it again. Your goal is a mix that nobody has to adjust.
To get there you need to separate two ideas that people often mix up. Peak level is how high the loudest instant of the signal goes. Loudness is how strong the sound feels to a listener over time. Two videos can have the same peak and feel very different, because one has steady, dense audio and the other has a few spikes over quiet speech.
This guide gives you a practical way to measure and set both. It avoids fixed numbers for individual platforms, because platforms apply their own loudness handling and change it over time. Instead, you will build habits that work everywhere: sensible peaks, consistent loudness, and a check on real devices.
Step-by-step
-
Understand what a meter is showing you
A standard track meter shows level in decibels, written dB. In most editors the scale tops out at 0 dB. That is the digital ceiling. Anything that tries to go above it is clipped, which means the sound is distorted. Almost all of your useful range sits below zero, so values are negative: -6 dB is louder than -20 dB.
Most basic meters show peak level, the highest instant. Some editors also include a loudness meter, often labeled in LUFS or LU. This measures perceived loudness across time. If your editor has one, learn where it is. If it does not, you can still do careful work with the peak meter and your ears.
-
Set dialogue first
Dialogue carries the information in most videos, so it sets the reference point for everything else. Play a section of clear speech and adjust the clip or track level so normal speech peaks around -6 dB to -3 dB on the meter. Many editors aim in that range, and it leaves room before clipping.
Do not aim for 0 dB. A line that peaks at the top has no room for a laugh, a raised voice, or a later effect. It is also more likely to clip when a platform converts the file. Keep some headroom. Headroom is the space between your loudest peak and the ceiling.
-
Match clips to each other
Once the main voice is set, check each clip against it. Scrub through the timeline and watch the meter. Interview answers, narration, voiceover pickups and on-camera lines may each have a different level. Adjust each clip so that normal speech lands in the same zone.
Use your ears as well as the meter. A bright voice can sound louder than a dull one at the same peak. If one speaker sounds consistently louder despite similar peaks, lower that clip a little. Consistency inside the video matters as much as the final number.
-
Place music and effects underneath
Music and sound effects support the picture. They should not compete with speech. Start with music at a low level while someone is talking, then listen. A common approach is to set music clearly below dialogue under speech, and bring it up in gaps or at the start and end where nothing is being said.
Use volume keyframes to lower the music before a person starts talking and raise it after they stop. Make the ramps smooth, over a half second or more, so the change does not draw attention. Your guide to balancing music and voice covers this in more depth. Use only music you have a license to use.
-
Control dynamic range with a light touch
Dynamic range is the gap between the quietest and the loudest parts. Wide dynamic range feels natural, but on a phone in a noisy place, quiet words can disappear. A compressor narrows the gap by turning down the loudest parts, and then you can raise the overall level.
Start gently. A small amount of compression on dialogue makes it easier to follow without sounding squeezed. If you hear pumping, where the background noise rises after each word, or if the voice sounds flat and lifeless, reduce the amount. A limiter is a stricter form of compression that stops peaks from crossing a set ceiling. Many editors put one on the master track as a safety net.
-
Check overall loudness, not just peaks
If you have a loudness meter, play the whole video and read the overall value at the end, often called integrated loudness. Compare this value across your videos. They do not need to match a platform's rule exactly, but they should sit close to each other so your audience does not have to adjust the volume between uploads.
Without a loudness meter, use a reference video. Pick one of your finished videos that sounds good on several devices. Play it, then play your new mix at the same system volume, switching back and forth. If the new one is obviously louder or quieter, adjust and test again. This is a crude method, but it works well for a series of videos.
-
Know that platforms may adjust your audio
Many video services apply their own loudness handling. Depending on the service, that can mean turning loud uploads down, leaving them as they are, or sometimes raising quiet ones. The rules differ and can change, so do not build your mix around one platform's guess.
The practical takeaway is simple. Do not try to win a loudness war by pushing everything to the ceiling. If a platform lowers loud videos, your squeezed mix will just sound flat and quiet after the drop. A clean, dynamic mix with solid peaks and consistent loudness holds up whatever the service does. Check the help page of each platform you publish to for the current guidance.
-
Test on real devices
Export your video and play the file on at least three things: headphones, laptop speakers and a phone speaker. Phone speakers have almost no bass, so music that seemed strong in headphones may sound thin and let the voice dominate. Laptop speakers may sound harsh on loud speech.
Listen to the quietest and loudest moments. Can you follow every word at a normal volume? Do any sections startle you? If so, return to the timeline and adjust. Listening on a phone at a moderate volume in a normal room is the closest test of how many viewers will actually watch.
-
Lock the mix before export
When the levels are right, put a limiter on the master track with its ceiling a little below zero, such as -1 dB, to catch stray peaks. Watch the meter for a final play-through. No red indicators should show.
Avoid last-minute normalizing of the whole project unless you check the result. Normalizing raises the loudest peak to a target, which does not equal consistent loudness. Save a note of the settings you used, so you can repeat them for the next video.
Examples
A talking-head YouTube video with background music
Narration peaks around -6 dB, and music sits clearly lower under it. In the intro and outro, where nobody speaks, the music rises. A limiter on the master catches an occasional spike. The creator compares the export with a previous upload at the same volume and finds it closely matches.
A tutorial recorded on two different days
The first session was recorded with a headset, the second with a desktop mic. The second sounds quieter and brighter. The editor raises its clip level until normal speech lands in the same zone as the first, then listens through headphones to make sure the second voice does not feel harsher. Viewers get one steady voice across the video.
A short clip for phones
Most viewers will watch on a phone without headphones. The editor applies mild compression to the voice, raises the music slightly so it is audible on a small speaker, and checks the result on an actual phone. The speech stays clear and the peaks stay below the ceiling.
Common mistakes
- Mixing peaks up to 0 dB. A signal sitting at the ceiling has no headroom and is easy to clip during export or playback conversion. Leave a few decibels of space.
- Confusing peak level with loudness. A peak meter tells you the highest instant, not how loud the video feels. Two mixes with the same peak can sound very different in loudness.
- Over-compressing to sound loud. Heavy compression and limiting make everything the same level. The mix feels tiring and may be turned down by a platform anyway.
- Mixing only in headphones. Headphones are forgiving. A phone speaker or laptop shows if the voice is buried or the music is thin. Test on more than one device.
- Chasing a number from the internet. Loudness advice is often outdated or tied to one service. Use sensible peaks, keep videos consistent with each other, and check each platform's current help page.
- Forgetting that music changes the balance. A track that feels quiet alone can cover speech. Always set music while the dialogue is playing.
Best practices
- Keep one reference video you like, and compare new mixes to it at the same system volume.
- Set your computer speaker volume once and leave it, so your judgments stay consistent.
- Put a limiter on the master track as a safety net rather than as a way to make things louder.
- Listen to the first thirty seconds and the last thirty seconds on a phone before you publish.
- Rest your ears. After an hour of mixing, a break changes what you hear.
Key takeaways
- Peak level and loudness are different measurements.
- Aim for dialogue peaking roughly between -6 dB and -3 dB, and keep headroom.
- Make loudness consistent between your videos.
- Platforms apply their own loudness handling, so avoid squashing your mix.
- Test the mix on headphones, a laptop and a phone.