Updated October 2026 · Editing audio well
Edit music to fit speech by timing beds to the rhythm of the talk, placing stings at the right points and shaping endings so they finish with the speech rather than fighting it. The music should support the voice, and it should change where the speech changes, not run on regardless of the words.
Music in a podcast or video is there to support the speech. It can set a mood, mark a transition or give the listener a moment to breathe. When it is edited well, the listener barely notices it. When it is edited badly, it competes with the voice, arrives at the wrong moment or ends in an awkward place. Fitting music to speech is therefore a matter of timing, level and sensitivity to what the words are doing.
This post covers the three main jobs: timing a bed under speech, placing stings and shaping endings. It assumes the music is licensed for your use, which is a separate check covered in earlier posts in this series.
A bed is music that plays under speech for a longer period. Time it to the rhythm of the talk, so it rises slightly in the pauses and falls back when someone speaks. Ducking can do this automatically, but check the result by ear. The bed should never compete with the consonants or the key words, so keep it well below the voice.
A sting is a short musical phrase used to mark a change, such as the start of a segment or a dramatic moment. Place it at the exact point of change, not a moment early or late. A sting that lands on a pause feels deliberate, while one that cuts across a word feels careless. Test the placement with the voice playing, since the sting and the speech can mask each other.
Music endings are where many edits go wrong. A fade out that ends in the middle of a phrase feels unfinished, and a sudden stop feels abrupt. Shape the ending to finish with the speech: let the music fall away as the final sentence lands, or hold a final chord for a beat after the last word. The ending should feel like a decision, not an accident.
Music and speech should share a sense of energy. A calm passage with a driving track can feel mismatched, and an excited talk with a slow bed can drag. Choose music that suits the mood of the section, and adjust its level so it fits the energy of the voice.
Most music in speech should sit well below the voice, so the words are always clear. A simple test is to listen to the speech alone, then to the speech with the music. The music should add atmosphere without changing how easy the speech is to follow. If you find yourself straining to hear a word, lower the music.
Music often sounds fuller on headphones than on a phone, where the low end is weaker and the voice can be lost. Check the balance on a phone speaker, and listen for any section where the music masks the speech. The ducking guide explains the automatic method, and the check on a phone catches what automation misses.
A music bed with its own vocal competes with the speaker, even at a low level. Choose instrumental versions for speech-heavy sections, or use a bed with the vocal removed where you have the right to do so. If the only version has a vocal, keep it well below the speech and place the speech where the music is most open.
When you shorten a track, cut at phrase boundaries in the music, not in the middle of a chord change. Listen for places where the music rises just as the speaker starts, and move the cut so the two do not compete. A well-chosen cut makes the edit feel like a musical decision rather than a patch.
Music editing depends on taste and genre, and there is no single right way to time a sting or shape an ending. Some styles call for a dramatic stop, and others for a long fade. Trust your ear, and use the principles as guides.
Editing music is only half the job. Every track must be licensed for the way it is used, including any edit. The intro music rules post sets out the basic checks. For the voice side of the balance, the Vocal Checker shows clipping and level problems in a recording you own, and the Tunedzy Go page covers cleaning phone recordings.
Spot clipping, noise, rumble and level problems. Runs in your browser, nothing is uploaded.
Well below, so every word stays clear. Lower it further if you strain to hear the speech.
At the exact point of change, tested with the voice playing.
Let it finish with the final sentence, or hold a final chord briefly. Avoid stopping mid-phrase.
It helps, but check the result by ear and on a phone, since automation can miss details.
