Give the voice room. Let music support it.

Start with the narration.

Add your recording under Audio and music → Primary audio, or create and review a short AI narration. Set the scene lengths around the recording. Listen for rushed endings and empty stretches before adding another sound.

Import or create timed captions after setting its source trim and video entrance. Replacing, removing or trimming that recording clears the old captions; Undo brings them back. Changing the recording’s video entrance keeps their times fixed: inspect their alignment deliberately. Adding or changing background music keeps them.

Add a track you have permission to use.

1. In Background music, confirm your rights and choose Upload music. Use an MP3 or PCM WAV up to 6 MB and ten minutes.
2. To reuse a file, open Reuse workspace media, filter to Audio, preview the file, confirm permission, and choose Use as background music.
3. Music starts at 15% volume. Treat that as a starting point: recordings differ in loudness, so listen and adjust.
4. Use Skip source intro (seconds) to skip a lead-in inside the file. Use Background music enters video at (seconds) to delay its first remaining sample. For example, skip 0.5 seconds of the file and enter at 1.2 seconds in the video; the skipped intro stays skipped and the first 1.2 video seconds have no music from this track. Primary audio has its own entrance control.
5. Set a Fade in or Fade out for either track, from zero to ten seconds. Zero leaves that end unchanged.
6. For a short backing track, select Repeat music to the last scene. Each repeat starts at your chosen source offset; an intro you skipped stays skipped.
7. Play the video with both tracks, save, and export the version you reviewed.

Place the entrance before the last frame.

Use tenths of a second from zero to just before the video ends. A track plays silence before its entrance, then the trimmed source. Choose Start with the video to remove the delay; Undo restores it. Moving primary audio keeps caption words and times, so listen and correct their alignment. Saved transcript candidates match the recording, source trim, entrance and video length; a changed entrance prevents applying an earlier candidate accidentally.

A caption-passage trim rebases each entrance and advances source trim only for audio already playing; original files stay intact. Tracks entering after the retained passage are removed from that new edit. Restore removed video timing before changing a placement that was used for word cuts. If placement is temporarily unavailable, its saved entrance stays visible: start both tracks with the video and save a new version, or try again later.

Choose a pace for each track.

Use Primary audio playback speed or Background music playback speed to choose 0.5×, 0.75×, 1×, 1.25×, 1.5× or 2×. One is the original pace. Each track keeps its pitch and original file; processing may change the sound, so listen to the finished MP4. A slower track lasts longer; a faster one ends sooner. Scene lengths and captions do not move automatically.

Source trim stays in original-file seconds. Entrances and fades stay on the video clock; repeating music follows its selected speed. A transcript candidate keeps the speed of its saved recording, with cue and word times projected onto that video; changing pace prevents applying an earlier candidate accidentally. A passage trim advances source trim by the amount actually played. Restore word cuts before changing a bound speed. Choose Use original speed to reset either track, then Undo to restore it. If this capability is unavailable, the saved choice stays visible; reset both tracks and save a new version, or try again later.

Find a cleaner starting point.

Choose Load waveform under either audio track to see its source peaks and quieter sections. The highlighted range follows the original source trim, playback speed and time remaining after the video entrance. If the source ends before the video, repeated music highlights the remaining source that will loop; primary audio plays once. Use the source trim control to move that range, then listen to the preview.

The waveform is decoded in your browser from the already loaded private audio, and the original stays unchanged. It shows source peaks, not the final volume, fades or mix limiting. If your browser cannot decode the waveform, the editor keeps the track and explains the recovery; use playback to choose a start or reload and retry. Loading a waveform does not save a video version.

Review scene changes against a steady rhythm.

Check a local tempo suggestion or supply your own BPM and first beat, then review nearby scene changes against that grid. Estimates need a steady track from 6 to 120 seconds; listen before using one. Keep repeat off and use cuts between scenes. The tool accounts for source trim, playback speed and the music entrance, then proposes one undoable edit while preserving total video duration and existing caption times. It does not track tempo changes or judge the musical result.

Let the music fall beneath your recording.

Choose Lower music under primary audio → Gentle or Strong. Ducking responds to the primary track’s sound level after its volume and fades; silence lets the music return. Gentle keeps more of the backing track, while Strong lowers it further. Start with a sensible music volume, then listen across quiet and louder passages. This is level-based ducking, not speech detection or noise removal.

The browser prepares a bounded approximation from your already loaded private audio without using a microphone. If preparation fails, it explains that preview plays without ducking and offers a retry; the saved export setting remains. Listen to the exported MP4 for the final balance. Without primary audio, the choice stays saved but has no effect. Choose Off to return to the ordinary two-track mix; Undo restores your previous setting.

Place separate sound effects.

Watch a six-second original mix and download its scenes, synthetic tones, caption file and exact settings to try three independent effects beside primary audio and music.

Open Sound effects → Choose sound effect from library. Upload your own MP3 or PCM WAV under Reuse workspace media if needed. Select and preview an audio file, confirm your permission for this video, and choose Use as sound effect. It starts at 30% volume for one second, with autosave off. Your existing narration, music and captions remain.

Add up to three effects. Give each a useful name, set its Video start and Duration in tenths of a second, then choose Skip source intro independently. The source must cover the whole duration. For example, start at 1.5 seconds, play for 1.2 seconds, and skip 0.5 seconds inside the original file: this effect plays video time 1.5–2.7 from source time 0.5–1.7. Each effect plays once at original speed.

Set volume and fades per effect. Fade-in begins at its entrance; fade-out ends at that effect’s window, rather than the last scene. Overlapping fades multiply. Effects can overlap each other and both main tracks; music ducking lowers only the background music. Listen across every overlap. Reusing the same file gives each placement an independent clock and counts its bytes once.

In Video timeline, select an effect to pause at its start. Drag its block to move the fixed window, drag Start/End to resize it, or use Edit selected timing for exact times. Review the proposed window before Apply block timing; this makes one undoable draft edit. The source start stays fixed and captions do not move. Preview, save and check the exported MP4.

Removing an effect retains its original file, narration and captions; Undo restores the placement. If the capability is unavailable, saved effects stay visible and can be removed before saving a new version. Effects support up to 12 scenes and 60 seconds at 720p, or 30 seconds at 1080p; each lasts 0.1–30 seconds. A caption-passage trim retains only intersecting sounds and advances source offsets; set their fades to zero first so fades cannot restart at a different time. Restore word cuts before changing a bound effect’s source or window.

Check more than the first sentence.

Use selected scene retiming to review several scene lengths together. Recordings and timed captions keep their video timestamps, so check alignment across the moved boundaries.

Listen across scene changes and through the final frame. A short track ends and leaves silence unless you enable repeat for background music. Narration never repeats. Music repeats without a crossfade at the seam, so choose an offset that sounds natural when the end meets it again. If either track is longer than the video, it stops at the last scene.

Fade-in begins at each track’s entrance; fade-out ends with the video. They do not restart on music repeats. A fade longer than the time remaining after the entrance is shortened to that time. Fade in and fade out multiply where they overlap. If a recording ends before the final fade, it leaves silence; lengthen it or shorten the video to soften that ending.

Reduce music when a quiet word becomes difficult to hear. You can set either track to zero while comparing the mix, then restore its volume. Watch the downloaded MP4 as your final check: exports limit loud mixed peaks, so a very loud mix can sound different from the browser preview.

Draw primary-audio waveform bars in the video.

Watch an original five-second waveform example and download its source pulse, editable scenes and exact settings.

Open Export primary-audio waveform bars after loading your permitted recording. Set a whole-percent box and a solid color, then choose Review waveform bars. Analysis stays in this browser and does not upload another file. Choose Inspect waveform preview, seek through quiet and audible passages, and listen before confirming the placement, source, sound and timing. Apply waveform bars changes one layer; one Undo restores it. Autosave remains off. Preview and Save video as a new version.

The 32 bars follow the primary recording’s source trim, video entrance, volume and fades. Silence leaves their box clear. Music and sound effects do not drive them. The bars sit above picture and captions, so leave useful wording unobstructed. They show short-window peak amplitude, not frequency, speech meaning or an automatically normalized loudness measure.

Use original-speed primary audio, no removed-word timeline, and at most 12 scenes and 30 seconds at 720p or 15 seconds at 1080p. For other timing, remove the layer first; your audio stays intact. After reloading a saved waveform video, open its controls and choose Load waveform preview before playing. Browser decoding may differ from the export; review the completed MP4, whose bars come from native filtered primary audio. If exports are unavailable, the saved setting is retained; Remove waveform bars and save another version to recover.

Inspect the finished mix.

After exporting an MP4, choose Review export → Load exported video. Open Inspect exported audio, then choose Load exported audio waveform. This graph shows the finished file’s combined soundtrack after source trim, track entrances, speed, volume, fades, repeat, ducking and encoding. It is separate from the original-track waveform in the editor.

Use Export playback position to move to a quiet passage or track entrance; seeking pauses the player. Press Play to listen from there. The position indicator follows the video, and Hide exported waveform lets you close the graph without changing the file or decision. Private guest reviewers can use the same inspection.

Analysis happens in your browser from the verified file already loaded for review. It does not play, upload or modify that file. Waveforms support self-contained H.264/AAC MP4 exports up to 6 MB and two minutes, with at most two audio channels. A silent MP4 gets an explicit no-audio result. Larger or unsupported files keep normal playback and review. The 8 kHz amplitude view does not normalize peaks and can miss fast details; it does not certify loudness, clipping or usage rights. Listen to the whole export before approving it. GIF exports are silent and have no audio inspection.

Measure loudness and peak headroom.

In Inspect exported audio, choose Measure exported audio to inspect the verified finished file. The three readings are Integrated loudness in LUFS, Estimated true peak in dBTP, and Decoded peak in dBFS. A peak close to or above zero deserves another listen and a check against your destination’s delivery requirements. A lower sample peak can still hide a higher peak between samples.

Measurements use the combined encoded soundtrack, including the main tracks, any sound effects and their processing. They stay in the browser, do not play or upload sound, and leave the file, source volumes and review decision unchanged. Choose Cancel audio measurements while analysis runs; the file remains available for playback, and you can retry.

To change the mix, return to Audio and music, adjust the track volumes, save and export a new version. Measure that new file and listen through it. No automatic loudness target or normalization is imposed: different destinations have different requirements.

The bounded tool decodes at 48 kHz, uses K-weighted gated loudness and four-times peak interpolation following the ITU measurement algorithm. Browser decoding or resampling can affect readings. This is not a certified broadcast meter, a loudness-range measurement, or a guarantee about distortion, intelligibility or rights. A silent file has no readings; very quiet audio may not pass the loudness gate. Analysis supports the same verified self-contained H.264/AAC MP4 files up to 6 MB and two minutes, and stops after 30 seconds. Larger or unsupported exports keep normal playback and review.

Know the current limits.

The studio supports one primary track, one background track and up to three independently timed sound effects, with separate volume, source offsets and fades, plus optional background-music repeat and level-based ducking beneath primary audio. It does not yet provide seam crossfades, changing-tempo beat tracking, a licensed music catalog or music generation. Footage’s original audio stays muted. All source media together must fit 12 MB per video; early-access MP4 exports run at 30 frames per second, up to two minutes with a 720-pixel short edge or one minute with a 1080-pixel short edge when that option is enabled.

Use music you own or a license that covers your intended destination and campaign. A file being available in the workspace does not establish those permissions. Removing it from the video keeps the reusable original in the private library.