Audio lanes work like video lanes: stack as many as the mix needs, and let auto-ducking keep the music out of the voice's way.
Lay out the lanes Drop music, voiceover and effects on separate lanes (+ adds more). Waveforms make alignment against the picture easy.
Set levels Each clip has its own volume; each lane has a master. Start with music around −12 dB relative to voice and adjust to taste.
Turn on ducking Flag the music clip as Duck in the Audio tab. It dips whenever the voice lanes carry speech and recovers in the pauses, in preview and export alike.
Shape the edges Per-clip fade in/out handles smooth entrances and exits; overlap two fades for a crossfade. Volume keyframes cover anything custom.
Clean up on export The Cleanup section can normalise loudness and reduce noise on rough recordings at render time. EQ, pitch, voice presets, echo and room reverb preview live and render the same way.
Tips
A video clip's embedded sound can be split onto its own lane to edit picture and audio separately. The two stay linked.
Remove silences (in the clip's Audio tab) scans for quiet stretches and jump-cuts them in one click. Every cut ripples, and a single undo restores them all.
Remove filler words sits next to it: a transcription-driven scan finds every "um" and "uh" so you can cut them the same way, unticking any you would rather keep.
Mute a lane to audition the mix without it; solo is one click away for the opposite.
Record voiceovers in the browser: the recorder is in the Assets panel.