
A spectral selection does not make ordinary Delete frequency-specific. Audacity documents that ordinary time edits ignore the frequency bounds. Undo an unintended time deletion before continuing, then choose a documented spectral effect if only a frequency band should change.
Key Takeaways
- A colored frequency box does not change every command's scope.
- Preserve the recording before destructive processing.
- Check the selected time, frequency bounds and intended effect separately.
Distinguish a time edit from a spectral process
The Spectral Selection manual defines a selection with both time and frequency ranges. Ordinary Cut, Delete and Trim remain available, but do not honor the frequency restriction. A waveform passage removed from the timeline is therefore not the same operation as attenuating a narrow whistle during that passage.
For a proposed narration, imagine a short tonal intrusion behind one word. The editor draws a narrow band in Spectrogram view, then presses Delete. If the entire word disappears or timing changes, stop and undo. Do not compensate by dragging later clips until the film seems aligned; first restore the original temporal structure.
Build the intended selection explicitly
Switch the track to Spectrogram and enable spectral selection through its settings where required. Define the affected time interval and the frequency limits. Inspect those limits with the Spectral Selection Toolbar rather than treating a tiny visual box as an exact measurement.
Display resolution affects how easily a tone can be located. The spectrogram window guide covers that tradeoff, but a clearer display does not change an ordinary editing command into a frequency-aware one.
Select a small disposable passage first. Preserve enough surrounding speech to hear whether a change damages the voice, and keep an unchanged source for comparison. A correctly selected band can still contain wanted consonants or harmonics.
Choose the documented effect rather than the key name
The Spectral Delete instructions give Effect > Spectral Tools > Spectral Delete. The command works directly without a parameter dialog. Check the actual release's menu before running it; an effect with a similar name is not proof of matching behavior.
That help also warns of technical limitations at very low frequencies. Do not use its example range as a universal prescription for hum or a microphone pop. The local plosive-repair guide addresses a different low-frequency problem. A narrow-notch approach may suit a stable hum, with its own tradeoffs.
Verify the unaffected content
For the proposed whistle check, compare duration and a recognizable word before and after the affected interval. Then compare the voice during the processed passage at a useful listening level. Success requires retained meaning and timing, not simply the disappearance of a bright line from the display.
Export a short proof only after the project comparison is satisfactory. Keep the source revision, selected time and frequency interval, effect name and observations in the review note. If the result thins the voice or introduces another defect, abandon the candidate and reassess the source rather than repeatedly deleting more bands. This is a proposed procedure, not a spectral repair performed on sample audio.
Related tasks
- Why Does an Audacity Spectrogram Show Clear Hum but Blur a Short Click?
- How Can I Reduce a Microphone Pop Without Thinning the Whole Voice?
- How Do I Remove Electrical Hum Without Thinning the Whole Voice?
Sources and scope
- Audacity Spectral Selection and Editing: Time-edit versus spectral-effect scope.
- Audacity Spectral Delete: Specific spectral deletion route and limits.
Documentation-based procedure; example is proposed, not tested.
Prepared by Imagild Editorial, the company editorial team responsible for these guides. AI assisted research and drafting. This is documentation-based guidance with proposed examples, not a claim of hands-on testing or an independent product review. Report a correction.
Cover: AI-generated conceptual editorial illustration; not a real application screenshot, observed output or before-and-after test.