More Clean-Up of Digitized Album
Last Updated: 03/20/2025
Clean-Up Overview
After the previously discussed minimal cleanup of the captured audio, the next step is to run the captured audio recording through the relevant Audacity tools to remove unwanted noise. Since every album has different noise, click, and pop issues, each capture has a unique and iterative tool flow using Audacity's five or six noise removal and repair tools.
This process can result in digitized recordings of well-used albums sounding better than the album itself. However, unless the vinyl album is in great shape, it usually sounds worse than a digital copy or CD.
​
As mentioned in other pages, the Audacity tools that are used and whose usage is described on this and subsequent pages are as follows:
​​​​
-
High Pass Filter
-
Click Removal
-
Repair (previously described here)
-
Noise Reduction
-
Generate Silence/Delete​​​
​
The order in which some tools are used can affect the result. For example, the Noise Reduction tool needs noise samples (with no other audio); it should be run before Generate Silence. Conversely, the Repair tool can be used anytime during the cleanup process.
​​​​
A note of caution: Depending on their settings, these tools can negatively impact the recording audibly. The settings for these tools are a tradeoff between reducing an audible problem (e.g., noise or clicks) and worsening the intended content. This tradeoff is particularly true of Click Removal and Noise Reduction tools.
​
High Pass Filter
Figure 1. High Pass Filter Menu and Dialog Box
The High Pass Filter tool removes frequencies below a specified frequency. When digitizing a Vinyl record, this filter can remove low frequencies that may be added to a recording due to a warped record. To use this tool, first select the portion of the recording to which to apply the filter; it makes sense to apply it to the whole recording. Then, select the tool as shown in Figure 1.
​
Since the frequency of the lowest note of a piano is about 27.5Hz, and the lowest note of a standard-tuned 5-string bass guitar is about 41Hz, 30Hz is a reasonable setting for the high-pass filter's cutoff frequency without impacting the recording. The rolloff should be relatively steep to remove any energy below the cutoff; thus, 24dB is a good setting.
​
Several tutorials suggest using this tool (for example, here in step 8), but I have not encountered warping that noticeably impacts the recording, so I don't generally use it. Also, some phono preamps (including the ART Phono Plus) have built-in high-pass (aka low-cut) filters that can be used instead.
​
Click Removal
Figure 2. Click Removal and Dialog Box
​Audacity's Click Removal tool is not as precise as the Repair tool, but it is more automated. This tool scans the selected recording portion, searching for specified height and width spikes. When such spikes are found, they are attenuated by likely interpolating between the start and end of the detected spike. In my experience, this tool can significantly improve audio quality, but it typically does not eliminate all clicks and pops entirely. This issue is more prominent in quieter passages, but the reduction is sufficient in louder sections to render most (but not all) clicks inaudible.
​
The steps to using the tool are as follows:
-
The desired portion of the recording is selected before selecting the tool
-
The Click Removal tool is selected, as shown in Figure 2.
-
Once the Click Removal dialog box is open, the user adjusts the spike detection threshold and width.
-
Depending on these settings, the spike detector may also mistakenly identify musical transients as spikes. Therefore, a trial-and-error method using the Undo command can be used to find the best results.
​
Figure 3 presents two examples of using Click Removal on a noisy or scratched recording. They demonstrate relatively typical usage over a few seconds of audio. The spikes, highlighted by yellow arrows, are visible on the right but have been reduced or removed on the left.
​
Figure 4 shows a zoomed-in view of a different situation where Click Removal did a good job removing a spike.
​​
Figure 3. Example Results when using Click Removal
Figure 4. Another example of Click Removal Working well.
Not depicted in the previous examples is that the algorithm seems more effective at finding and removing spikes that exceed the amplitude of the adjacent waveform samples, but is less capable of addressing a spike that is buried within the waveform.
​
Figure 5 illustrates an example of a spike/glitch that the Click Removal tool did not remove. This glitch is a smaller amplitude, so perhaps the settings were not sensitive enough (though I attempted to adjust them). Additionally, it is possible that because this spike is embedded within the amplitude of the surrounding samples, the algorithm fails to detect it. In either case, the glitch was audible and can be fixed using the Repair function, as demonstrated on the right side of Figure 5.
​​
Figure 6 illustrates another, perhaps atypical, glitchy noise event. This one sounds like a brief burst of static. As shown in the figure, this noise resembles a few milliseconds of random spikes. Applying Click Removal helped to reduce the peaks of some spikes but, not surprisingly, did not fix the smaller spikes, even when employing very sensitive settings.
​
Figure 5. A Spike Where Click Remove is Ineffective
Figure 6. Menu Save Project and Export Audio and Their Dialog Boxes