How to Add Music to a Video With AI: Generate, Then Mix

You add music to a video by generating an instrumental in the music tab, then placing it on one of the two audio tracks in the Timeline Editor and setting its volume and fades against your voice. MiniMax Music 2.6 costs 12 credits per song, and ACE-Step costs 6 credits per generation in either its Base or Turbo version. The clip and the audio are mixed and rendered together, with the render fee shown before you confirm.

Updated October 2026.

Pick the mood first: tempo, energy and platform

Music decides the pace the viewer feels before they understand the content. A fast track over a calm product shot makes the video feel urgent; a slow track over a fast edit makes it feel considered. Pick the mood before you write the prompt, because vague prompts produce generic music.

• Tempo: roughly 70 to 90 beats per minute feels calm and explanatory, 100 to 120 feels lively, and 130 and above feels energetic or athletic.
• Energy: decide whether the track should sit under a voice or carry a silent section. A track with a strong melody competes with speech; a sparse bed does not.
• Platform: a 15-second ad needs a track that establishes itself in two bars. A 90-second explainer can afford an intro.

Write the mood as three or four words before you open the tool: “warm, steady, acoustic” or “tense, minimal, electronic”. That phrase becomes the core of the style prompt.

Generate an instrumental: style prompt, length, no vocals

Open the music tab. Two model families are available: MiniMax Music 2.6, which produces full songs with vocals when you give it lyrics, and ACE-Step, which is built for shorter instrumental generations.

For background music, choose instrumental mode. On MiniMax Music 2.6 that is a dedicated toggle; with it on, the lyrics field is not needed and the output is a backing track. On ACE-Step, set the length you need. The style prompt is where the mood phrase goes, alongside genre and instrumentation: “warm acoustic guitar, soft percussion, steady, no vocals”.

Pricing is per generation, not per minute. MiniMax Music 2.6 is 12 credits per song and each ACE-Step generation is 6 credits, whether you pick Base or Turbo. The price is shown before you generate, and a failed generation is refunded. Generate two or three short options rather than one long one: a 30-second bed is easier to judge than a three-minute track, and the unused options cost little.

Listen for one thing before you use a track: does it have a clear loop point or a natural ending? A track that fades out works under a fade; a track that stops abruptly needs a cut.

Open the Timeline Editor and add the clip and the audio

Go to the Timeline Editor and create a project in the shape you need: 9:16 for Reels, TikTok and Shorts, 16:9 for a landing page, or 1:1 for a square placement. Add your video to the single video track, either from My Creations or as an upload.

Then add the audio. Two routes work. A generated track appears in your creations and can be selected from the library. A track you already have can be uploaded; the editor accepts common audio formats including MP3, up to 25 MB per file. Drop it onto one of the two audio tracks.

The editor is a sequence editor, so the music sits alongside the video rather than inside it. That is what makes the mix adjustable: you can move the start point, change the level and fade it independently of the picture.

Set where the music starts and ends, with volume and fades

Place the music so it begins on the first frame that matters. A track that starts before the picture can feel like it is waiting for the video. Trim the audio item to the section you want, then set its start time on the timeline.

Each audio item has its own volume, fade in and fade out. Use them together:

• Fade in over the first second so the track does not arrive at full volume on the first frame.
• Lower the music under speech. If the voice-over and the music share the same track, split the audio and lower the music item where the voice runs.
• Fade out at the end rather than cutting the track, unless the video ends on a hard cut that the music should match.

With two audio tracks, the cleanest setup is voice on one and music on the other. That keeps the levels independent all the way to the render, and it means a change to the music never touches the voice. If you need a voice-over first, the AI voice-over guide covers generating one.

Add sound effects at the cuts

Music carries the mood; sound effects carry the action. A whoosh on a transition, a click on a button press or a riser before a reveal makes an edit feel intentional. Place them on the second audio track, or as additional items on the same track, at the exact frame of the cut.

The sound effects tab generates an effect from a text description, which is useful when you need a specific sound that is not in a stock library. Our guides to AI sound effects and the best sound effect generators cover prompts and alternatives.

Keep effects short and sparse. One accent per cut is plenty; an effect on every transition turns the soundtrack into noise. If a generated video already has its own audio, decide whether it stays: native audio and a music bed can work together, but the levels need balancing.

Rights: your own audio, generated audio and what to check

Three kinds of audio end up in a project, and they carry different questions.

• Your own recording: you own the master, but the people in it and any music played in the room may have separate rights. Get consent for voices, and do not assume a track playing in the background is yours to use.
• Generated audio: the terms of the model you used govern how the output can be used. Read the terms for the specific model before you monetise a video, and keep a record of which model produced which track.
• Third-party tracks: a track you did not create and did not generate needs a licence. Do not assume that a track found online is free to use in an ad.

VdoBloom does not grant rights it does not hold, and this is not legal advice. Check the model terms and, for a commercial campaign, check the rules for your platform and your market.

Check the mix on a phone speaker

Most viewers will hear the video through a phone speaker, so that is where the mix should be judged. Export, play it once at low volume, then once at normal volume. Speech should stay intelligible at both; music should support it without covering it.

Watch the levels too. If the waveform clips at the loudest moment, the render will distort there. Leave headroom by lowering the music rather than raising the voice. And check the ending: a fade that sounds smooth on headphones can feel abrupt on a small speaker. For the assembly side, see stitching clips in the Timeline Editor.

Rendering costs a flat 10-credit fee, shown in the editor before you confirm, and failed renders are refunded. Credit packs are one-time purchases that never expire, and the options are on the pricing page.

Frequently asked questions

Can I generate music without vocals?

Yes. MiniMax Music 2.6 has an instrumental toggle, and ACE-Step is built for instrumental generation. Describe the style and leave the lyrics field out.

How much does a generated track cost?

MiniMax Music 2.6 is 12 credits per song. ACE-Step Base and Turbo are 6 credits per generation. The price is shown before you generate.

Can I use my own music file?

Yes. Upload it to the editor, which accepts common audio formats including MP3 up to 25 MB per file.

Can the music and a voice-over play together?

Yes. The editor has two audio tracks, each with its own volume and fades, so voice and music stay independent to the render.

Is generated music cleared for commercial use?

That depends on the model terms. Read the terms for the model you used before monetising, and check your platform’s rules. This is not legal advice.

Open the music tab and generate a short instrumental for the mood you wrote down, then take it into the Timeline Editor and place it under your clip. New accounts start with 10 free credits and no card required, which covers an ACE-Step generation or most of a MiniMax song, so you can hear a track against your own footage before buying more.