Music makes a video feel finished, and it also ruins a lot of them. Too loud, and nobody can hear the speaker. Wrong mood, and the whole thing feels off. Finding a track you may use is its own chore. Superframes can generate a track from a short description and place it under your video, and it lowers the music whenever someone is speaking. This guide covers how to ask for the right track, how to set it up, and what to check before you export.
Ask what the music is doing. For a talking video, it is a bed: it fills the silence and sets a mood, and nobody should notice it. For a montage with no speech, it is the main thing, and the cuts can follow it. For an intro, it is a short signature. A bed wants a steady pace, few changes and no strong melody. A track with a clear tune competes with the voice even when it is quiet, because the ear follows the tune. That is the most common mistake in talking videos.
Open the Generate window, pick the Music tab and describe what you want in one or two sentences. Name the mood, the pace, the instruments and the use. "Calm, warm instrumental for a talking video, soft piano and light pads, slow, no vocals, no big drops" is a good prompt. "Chill music" gets you whatever the average of chill is. If the first track is not right, change one thing and try again: the pace, an instrument, the mood. A failed generation costs nothing, but a track that came out fine and is simply not what you wanted does cost credits, so look at the price on the button each time.
The finished track lands in your media. Put it on an audio lane under your video and trim it to the length of the video. End it with a fade so it does not stop dead. If the track is shorter than the video, either generate a longer one or place the track twice, and cut the second copy at a natural pause.
Fixed music volume is a compromise. Set low enough to hear the speaker in the loudest part, it vanishes in the gaps. Superframes handles this for you. Turn on the lowering for the music item and the editor finds where speech happens in your video, using the words it worked out on your device, and lowers the music for those stretches. It brings the music back up in the pauses, with a short ramp of about a quarter of a second in each direction so the change is not a jolt. This needs speech it can find. If the video has no recorded voice on the main lane, there is nothing to lower for, and the music stays at its set level.
Two levels matter: the music while someone speaks, and the music in the gaps. While someone speaks, the music should sit well below the voice, so that you only notice it when you listen for it. In the gaps, it can come up, but not to the point where the return is startling. Check with headphones and then with phone speakers, because most people watch on a phone. If you cannot follow every word with the music on, lower it more. Also, listen to the whole video once, from the start, with your eyes closed. Any place where your attention goes to the music instead of the words needs fixing.
A music track is 20 credits, and 100 credits is one US dollar. The price shows before anything is made. If the generation fails, is refused or is cancelled, it costs 0. When you generate, your prompt goes to our AI providers, and nothing else does. Your video files stay on your device, and the lowering of the music also runs on your device.
Make a quiet video louder or tame loud audio in your browser. Adjust volume in dB without uploading your clip.
Swap a video’s audio track with music or voiceover from another clip, in your browser. Video stays untouched and is never uploaded.
Add smooth fade-in and fade-out transitions in your browser. Professional open/close fades without uploading your footage.
Turn on the lowering for the music in Superframes. It finds where someone speaks and lowers the music for those parts, then brings it back in the pauses. Also ask for an instrumental with a steady pace and no strong melody.
A music track is 20 credits. The price shows before it is made, and a failed generation costs nothing.
No. The words in your video are worked out on your device, and the lowering runs there too. Only your prompt leaves your device when you generate the track.
Not for a video where someone speaks. Singing competes with speech for attention even at a low level. Ask for an instrumental.
Then there is nothing to lower for, and the music plays at the level you set. For a montage, that is usually what you want.