How to choose background music for a video

To choose background music for a video, decide the mood first, match the tempo to how quickly the scenes change, check the licence covers where the video will be published, and set the level well below the voice. For a promo that usually means an instrumental track with a steady pulse; for a tutorial it often means a very quiet bed or nothing at all.

The order matters. Most people start by browsing a library until something sounds good, then discover it fights the voiceover, drags against the cuts, or cannot be used in an ad. Working from mood to tempo to licence to level avoids all three.

Mood first: decide what the viewer should feel

Write one or two words for the feeling the video should leave: lively, urgent, warm, calm, confident. That word eliminates most of a library before you listen to anything.

Take it from the brief, not from the product category. The restaurant opening video opens on fire and food under the line “Abrimos el viernes” and is narrated in an energetic voice, so it wants something warm and lively. A different restaurant announcing a quiet tasting menu would want something slower and more intimate, even though both are restaurant promos. The real estate listing video uses a warm voice and a light look to walk through a flat in Lisbon; its music should feel open and unhurried, not like a sale.

A useful check: play the track under the first scene with the sound of the voiceover and ask whether the two are telling the same story. A cheerful track under a serious message reads as careless, and a solemn track under an offer reads as a warning.

Tempo: match the pulse to the cuts

Tempo is how fast the music feels, and it should agree with how fast the video moves. Promo scenes are short. The restaurant video runs six scenes in 24 seconds, and the countdown offer video runs four scenes in 18, so each cut lands every three to five seconds. A slow, drifting track under that feels like the edit is rushing ahead of the music.

Tutorials move the other way. The expense claim how-to video holds most steps for 7 seconds so the viewer can follow each click. A fast track under that makes a calm walkthrough feel anxious.

There is research behind leaving music out of teaching videos altogether. In their 2003 review of multimedia learning, Mayer and Moreno describe a coherence effect: people understand a narrated explanation better when interesting but extraneous material is removed, and they name background music as one of those embellishments. Their suggested fix is weeding, which means cutting anything that does not help the viewer follow the explanation. For a promo the goal is feeling, so music earns its place. For a how-to the goal is a viewer who can repeat the steps, so the question to ask is whether the track helps that happen. If it does not, a clear voice with no music is the better choice, and the viewer rarely notices the absence.

Two more tempo checks. Avoid tracks with a big build or drop in the middle unless it lands on the scene that deserves it, because a crescendo under a checklist is distracting. And listen to the end: a track that finishes on a strong resolved note suits a call to action far better than one that simply stops.

Ducking: let the voice win

Ducking means lowering the music automatically whenever someone speaks, then bringing it back up in the gaps. It is the single biggest difference between a video that sounds produced and one that sounds like two files played at once.

WCAG gives a useful target. Its AAA success criterion 1.4.7, written for prerecorded audio that is mainly speech, asks for background sound at least 20 decibels below the speech, which it describes as about four times quieter. Video is not audio only, so this is a benchmark rather than a requirement, but it tells you how far apart voice and music should be when the words carry instructions.

Ducking only lowers the music relative to its own level. If the track starts too loud, a ducked track is still too loud. Set the base level first, then let ducking do the rest.

Licensing: know what you are allowed to publish

Music licensing is where good videos get blocked, muted or taken down. The safe assumption is that every commercial recording has owners, and more than one kind of right is involved. APRA AMCOS, which licenses music use in Australia, notes that businesses using music online, including in social media video, may need a licence, and that recording music into a video involves synchronisation rights cleared with the publisher.

Production music libraries exist to avoid that. YouTube’s Audio Library offers royalty free production music and sound effects inside YouTube Studio, marks which tracks need credit in the video description, and YouTube says only Audio Library content is known to it to be copyright safe. Other libraries set their own terms, and royalty free never means condition free: check whether paid advertising and every platform you plan to use are covered.

Detection is automatic. YouTube’s Content ID scans each upload, and when it finds a match the copyright owner can block the video, run ads on it or track its viewing. The match comes from the audio, so a description that credits the artist does not prevent it.

Volume: set it for a phone speaker

Most short video is heard through a phone speaker, which is small and loud in the middle frequencies, exactly where speech sits. Music that feels subtle on studio headphones can bury a voice on a phone.

Set the level with the voiceover playing, on a phone, at the volume people actually use. Then check three moments: the first line, because viewers decide in seconds; any line with a number, price or date; and the call to action. Creator style ads need the lowest level of all. In the fitness app UGC video, the point is that a person seems to be talking to you, and loud music turns that into an advert. More on that format is in what a UGC ad is.

A mood table to start from

The table below the article maps common video types to a mood, a tempo and a level, using the example videos on this site so each row can be heard in context. Treat it as a starting point: the brief decides, and a quiet tasting menu promo belongs in a different row than a grand opening.

Common mistakes

• Picking the track before writing the script. The words decide the mood; the track should follow.
• Vocals under narration. Two sets of words at once means neither is understood.
• Loud music in a tutorial. Instructions need attention, and music competes for it.
• Assuming royalty free means free for ads. Read the licence for commercial and paid use.
• Mixing on headphones only. Check the final mix on a phone speaker.
• Ignoring the ending. A track that cuts off mid phrase undermines the call to action. Pick one that resolves, or fade it out.
• The same track on every video. A regular viewer notices, and the brand starts to sound like a template.

Build it

The video builder has 13 music tracks, plus the option to upload a file or paste a track URL. The default music volume is 25 percent and the slider stops at 60 percent. Wherever narration plays, the music ducks to 22 percent of its set level, about 13 decibels down, ramping over 12 frames on each side. It ducks under narration only: audio kept from an uploaded video clip does not duck the music, so lower the music by hand in those scenes. The background music page describes the library and the ducking, how long a promo video should be covers scene pacing, and the steps for adding a track are in the tutorial on adding voiceover, captions and music.