How to Add Music to a Video (Manually, and the Faster Way with AI)

Before you start: three decisions that matter more than the software

• Mood and energy. Watch the video once without sound and note where the energy rises and falls. A talking-head explainer wants steady, low-key music; a travel montage wants movement and a clear build.
• Where the music should change. Most videos have two to five moments where the tone shifts: the hook, the first real point, a turn, the close. Music that changes with them feels edited; music that loops under everything feels like a playlist.
• The license. Decide up front whether the track can be used in a published (and possibly monetized) video. Fixing this after upload means re-exporting and re-uploading.

Method 1: on your phone (CapCut, iMovie, InShot, YouTube Create)

Phone editors are the fastest way to add one track to a short video. The workflow is almost identical across apps:

1. Import the video and tap Audio or Music.
2. Pick a track from the app's library (check its license tag) or import your own file.
3. Drag the clip so it starts where you want. Trim the end to the length of the video.
4. Lower the music volume to roughly 15–25% while someone is talking. Most apps have a Volume slider per clip.
5. If the app offers Fade in / Fade out, add about one second to each end.
6. Export at the same resolution as the original.

Limits: one track for the whole video, manual trimming for every scene change, and library tracks whose license often covers only that app's own platform.

Method 2: in a desktop editor (Premiere Pro, DaVinci Resolve, Final Cut, CapCut desktop)

Desktop editors give you full control: several tracks, cuts timed to the frame, keyframed volume and real ducking. The price is time.

1. Put the music on its own audio track below the dialogue track.
2. Cut the music at every scene change you noted earlier, and move or swap the segments so the energy matches.
3. Add a short crossfade (0.5–1 s) at every cut so the transitions are not audible.
4. Set the music level 18–24 dB below the voice while people speak. In Resolve and Premiere you can automate this with the ducking tools (Fairlight's Ducker, Premiere's Auto Ducking in Essential Sound).
5. Check loudness on export: around −14 LUFS integrated for YouTube, with the voice clearly dominant.
6. Keep the music file and a note of where it came from in the project folder, in case you ever need to prove the license.

Expect 30–90 minutes for a ten-minute video once you include searching for a track that works. Editors who do this daily usually keep a personal library of tracks they already trust.

Method 3: let AI score the video (what Scora does)

The third option skips the search and the trimming. You upload the video, and the AI analyses what is happening in it: the transcript, the pacing, the emotional shifts. It then segments the video and places music that matches each segment, with transitions at the moments that matter and the music automatically lowered under speech.

With Scora, the output is the scored video, the soundtrack as a separate MP3 (so you can drop it into your own editor), and a cue sheet showing what plays when. The music is either chosen from a curated, licensed library or generated for that specific video, which means it cannot already belong to someone else.

You still make the creative calls that matter: how loud the music sits, how many changes you want, and a theme or feel to steer the score. What you do not do is scrub through a timeline for an hour.

1. Create a free account at scora.fm. No card is needed, and your first video does not even require confirming your e-mail.
2. Upload the video (optionally with a subtitle file). Choose a scoring method: a curated-library score is free; AI-generated, continuous scoring is on paid plans, with a free trial on your first short video.
3. Set the music volume and how often it should change, or leave the defaults.
4. Download the scored video, or just the MP3 and the cue sheet if you prefer to finish in your own editor.

Volume: the rule most videos get wrong

If viewers notice the music while someone is talking, it is too loud. A safe range is music 18 to 24 dB below the dialogue during speech, rising to a comfortable level in sections without voice. Headphones exaggerate music; check at least once on a laptop speaker or a phone.

Ducking (lowering the music automatically when there is speech) is what makes professional videos sound professional. Do it with keyframes, with your editor's ducking tool, or let the AI handle it.

Copyright: what gets videos claimed, muted or demonetized

• Commercial songs will be detected by Content ID even at low volume or sped up. Expect a claim, sometimes a block.
• 'Free' tracks from random sites are often not free for published or monetized use, and some are registered in Content ID by third parties. Read the license, keep a copy of it.
• Royalty-free libraries are safe when the license explicitly covers online publishing and monetization. Keep the invoice or the license text.
• AI-generated or video-specific music cannot be in anyone's fingerprint database beforehand, which is one reason generated scores are becoming popular for channels that upload often.

Which method should you use?

• One short clip, one track, no dialogue: a phone app is enough.
• Interviews, tutorials, courses, podcasts with video: you need ducking and a few music changes. Desktop editor if you have the time, AI scoring if you do not.
• Weekly or daily uploads: the search-and-trim loop is where the hours go. Automating the score and finishing in your editor is the usual fix.

Frequently asked questions

How do I add music to a video for free?

Phone editors include a basic library, and Scora's free plan scores up to five videos a month from a curated library, up to ten minutes each, with no card required. In both cases check that the license covers how you will publish the video.

How loud should background music be in a video?

About 18 to 24 dB below the voice while someone speaks, and around −14 LUFS integrated for the whole mix on YouTube. If you notice the music during dialogue, lower it.

Can I add music to a video without it getting a copyright claim?

Yes, if the music is licensed for online publishing (keep the license) or was generated specifically for your video. Commercial songs and unlicensed 'free' tracks are the usual cause of claims.

Can AI add music that changes with the scenes?

Yes. Scora analyses the video's transcript and pacing, splits it into segments and places or generates music per segment, with the music lowered automatically under speech.