
A spectrogram setting that makes a steady low-frequency tone easy to distinguish can make a brief event harder to locate in time. In Audacity, increasing the FFT window size trades temporal resolution for frequency resolution. Changing the display does not itself repair or modify the sound.
Key Takeaways
- Match the spectrogram view to the event you need to inspect.
- A sharper-looking display is not proof of a better audio result.
- Verify any repair by listening before accepting it.
Name the diagnostic question first
Decide whether you are locating an event or inspecting its frequency pattern. A click between two words needs a precise time boundary. A sustained hum needs enough frequency detail to distinguish its line from nearby content. Trying to answer both questions with one view can create false confidence.
Audacity's Spectrogram Settings documentation describes the FFT window tradeoff and explains that zero padding changes interpolation rather than that fundamental resolution tradeoff. Use the settings to improve inspection, not as an audio enhancement preset.
Compare views of the same passage
- Preserve the unprocessed audio and select a short passage containing the problem.
- Note the current spectrogram settings before changing them.
- Compare a larger and smaller window while keeping the same passage visible.
- Use waveform view and playback to confirm the event's actual location.
- Only then select a repair method that matches the audible fault.
Do not change several audio effects while comparing display settings. Keeping inspection separate from processing makes it possible to tell why the apparent result changed.
Work through a mixed-noise example
Imagine a narration recording with a steady electrical tone and one small mouth click. Use the view that resolves the steady tone to inspect its frequency region. Then inspect the click with a setting that makes its timing easier to judge. The narrower-looking line in one view does not mean the recorded tone became narrower; the underlying samples have not changed.
After a repair, listen to the word before and after the edited area at an ordinary monitoring level. Compare consonants, pauses, and room tone. If a cleaner picture accompanies damaged speech, the repair has failed the listening test.
Decide what to deliver
An analysis screenshot can help explain a decision, but the recipient normally needs the approved sound file and a clear account of any processing. Keep an unprocessed source for future review rather than overwriting it because the display looks cleaner.
For separate treatment of a sustained electrical tone, read the narrow-notch hum guide. For local speech damage, the mouth-click comparison explains why preserving consonants matters. Neither article turns a spectrogram setting into a repair effect.
Related tasks
- How Do I Remove Electrical Hum Without Thinning the Whole Voice?
- How Do I Remove Mouth Clicks Without Damaging T and K Sounds?
Sources and scope
- Audacity Manual: Spectrogram Settings: Larger FFT windows improve frequency resolution and reduce temporal resolution; zero padding does not change that tradeoff; spectral selection supports time-frequency inspection.
Prepared by Imagild Editorial, the company editorial team responsible for these guides. AI assisted research and drafting. This is documentation-based guidance, not a claim of hands-on testing or independent review. Report a correction.
Cover: Pignatta / Pixabay, Content License. The city photograph is an illustrative source asset; it does not show this application or a test result.