Using the sound mixer and loudness meter in Video by Zaweba
View > Sound mixer in Video by Zaweba gives each track a fader, pan, mute, solo and peak meter, plus an EBU R128 loudness meter and fader automation.

Setting levels clip by clip works for a short piece. For anything with dialogue, music and effects on several tracks, it helps to see all the tracks side by side and move them like a desk. View > Sound mixer in Video by Zaweba opens exactly that.
A strip for every track
The mixer has a strip for each sound track. Each strip has a fader that works in decibels, a pan control to place the track left or right, mute and solo buttons, and a live peak meter that shows how loud the track is as it plays.
Mute and solo are the quickest way to find a problem. Hear a buzz somewhere in the mix? Solo each track in turn until you find the one it lives on. Want to judge the music on its own? Solo the music track and listen.
Peak meters show the loudest moments. Keep them out of the top of the scale on loud passages, because sound that hits the ceiling distorts.
The master strip and loudness
At the end of the row is a master strip for the whole mix, and it carries a loudness meter built to ITU-R BS.1770-4 and EBU R128, the standards broadcasters and streaming services use to judge loudness. It shows four readings.
Momentary loudness reacts to the last instant and jumps around. Short-term loudness smooths over a few seconds, which is good for judging a passage. Integrated loudness is the average over everything played so far, the number delivery specs usually ask for. True peak shows the highest point the sound reaches once it's converted back to an analogue signal, which can be a touch higher than the sample peak.
Peaks tell you about distortion; loudness tells you how loud the video feels. Two videos can peak at the same level and feel very different in loudness. The meter lets you compare your mix with what you're aiming for and keep a series consistent from episode to episode.
Automation: Read, Write, Latch and Touch
Moving a fader while the video plays is often the most natural way to ride a level, dipping music under a line or lifting a quiet voice. The mixer can record those moves as keyframes on the clip-volume fader, using four automation modes.
Read plays back the keyframes already there without recording. Write records your fader position for the whole time playback runs, replacing what was there. Latch starts recording when you touch the fader and keeps holding the new level after you let go. Touch records only while you're holding the fader and returns to the existing keyframes when you release it.
Touch is the forgiving one for small fixes: grab, adjust a line, let go. Latch suits a change that should stay, like lowering the music from a certain point onwards. Write is for a full pass when you want to perform the whole level from scratch.
A workable mixing routine
Start with the dialogue, because it carries the story. Set it at a comfortable level, then bring the music up underneath until it supports without competing. Effects and ambience go last. Watch the integrated loudness over a full play-through and adjust the master or the main tracks as needed.
Video by Zaweba's automatic ducking already lowers music under speech, so the mixer is for the fine adjustments on top of that, not for doing everything by hand. Keep the mixer open during your final listen, and fix anything you hear while it's fresh.