Why does Audacity Auto Duck leave my product-ad music unchanged while the voice is clearly audible?

What Auto Duck actually reads

Auto Duck has two roles. A target track is selected because its volume should be reduced; for a product advertisement, that is usually the music. The control track is unselected and supplies the signal that triggers the reduction; this is usually the voiceover. Audacity documents that the first unselected track underneath the selected track or tracks becomes the control track.

At least one selected track must be above the control track. Other tracks below the control track can also be modified if they are selected. The voice and music must also be synchronized. These rules explain why a clearly audible voice may have no effect: it may not be the control track Auto Duck analyzes, or the music may not be in the affected selection.

1. Duplicate the working project or use copies of the tracks.
2. Write down the track order from top to bottom.
3. Select only the music track or tracks intended to duck.
4. Deselect the intended voice control track and confirm it is the first unselected track underneath the selected music.
5. Confirm at least one selected music track is above that voice track and that the clips are synchronized.
6. Apply Auto Duck to the copy and listen around speech. Optional waveform inspection is a manual aid; it does not establish which control track was used.

Separate routing from timing

Do not change Duck Amount, Fade Down Length, Fade Up Length or Maximum Pause while the wrong control track is selected. Duck Amount is the volume reduction applied to selected tracks when a signal above the threshold is detected. Fade Down and Fade Up describe how quickly the music moves into and out of the reduced level. Maximum Pause prevents over-rapid returns; Audacity states that the volume will not ramp up unless the pause is at least as long as the fade-down length plus the fade-up length, regardless of the setting.

Auto Duck analyzes the control track and applies the effect to the selected tracks; verify the resulting audio on the copy. If routing is correct but music changes only briefly or returns later than expected, compare the speech gaps with the pause and fade lengths. Do not interpret a deliberately held duck as failed detection.

Illustrative exercise: two arrangements

Illustrative exercise, not a test performed here: use copies of a 12-second voiceover and a music bed. The timing is invented for the exercise. Arrange A from top to bottom as Music (selected), then Voice (unselected). Assume the fictional voice exceeds the chosen detection threshold during the phrase at 3–7 seconds. After applying the effect, listen for a reduction in music during that interval. Inspecting its waveform is an optional manual check, not proof by itself of which control track was used; verify the selected state and order separately.

Arrange B from top to bottom as Music (selected), Silent spacer (unselected, containing silence throughout the exercise), then Voice (unselected). The first unselected track underneath the selected music is now the silent spacer, not the voice. That silent control does not cross the detection threshold. If your test behaves as this hypothetical arrangement predicts, the music does not duck even though the voice on the lower track is audible. Restore the copy, move Voice directly underneath Music, select only Music and rerun. This compares control assignment while holding the music and voice recordings fixed.

The decision branch is simple: if playback confirms a music reduction during the voice phrase, continue to timing review; if it does not, stop and do not invent a threshold fix from this evidence. Recheck synchronization and the actual selected state, then seek documentation for the installed Audacity version before changing unsupported controls. Stop when the voice-triggered reduction occurs, the music returns at an acceptable point, and the approved original remains untouched.

Preserve and verify the result

Auto Duck analyzes the control track first and then applies the effect to selected tracks. Keep the pre-effect project or duplicated tracks available because the operation changes the target audio. Review the complete advertisement, including speech-free gaps and the final callout, rather than judging one waveform region.

Acceptance requires three observations: the intended voice is the documented control track, only the intended music tracks are selected targets, and the resulting music envelope lowers during speech and returns without an unacceptable delay. If any observation fails, restore the copy and return to the corresponding branch instead of repeatedly applying Auto Duck.

1. Label the copy with the track order and selected state used.
2. Listen through speech, speech gaps and the ending.
3. Compare the result with the untouched original.
4. Export or hand off only after the routing and audible envelope meet the project’s own review standard.

Auto Duck routing and envelope worksheet

These are illustrative exercise rows. Save a copy and record your own observations; the example is not an executed acceptance test.

Download review worksheet (CSV)