Click Repair Observations and Question/Suggestion

[GDGrass]

Having been using Vinyl Studio since 2016 and have performed quite a lot of click scanning and repair.  Not that I am an expert, but am continually learning how to do a better job at it as I go.
In searching for clicks, I noticed the following:

• A lot of recordings of records show the majority of clicks are found in the first few seconds of the music.  Apparently this is caused by the way the records are generally handled by holding the record along the recorded surface near the outer edges.  When this is the case, I usually perform an auto click repair only on the beginning of the recording and do manual searching on the balance of the recording to avoid introducing unwanted repairs on music that doesn't respond well with auto click repair.
• the right and left waveforms generally appear very similar in stereo recordings.  Of course mono tracks left and right waves are identical
• The clicks generally have a very sharp waveform spike but some music also produces a similar sharp waveform shapes too (This is especially true where the music waveform is very smooth like in soft/quiet areas)
• Many times the automatic scanning will pick up music waveforms where there is a sharp waveform change or spike and apply a correction
• Buy comparing the right and left waveforms at the correction location can help determine if indeed the correction applied by the auto scanning is indeed a click or possibly just a natural music spike or sharp change in the music waveform – I have found on music with waveforms that resemble these sharp waveforms, the auto click repair makes many correction on the music that can actually create a click.  By deleting the correction then clears the click.
• The waveform at very noticeable clicks usually show a large dB waveform departure from the surrounding music wave form which makes them easy to spot but sometimes missed by the auto correction process
• Listening to the general click location that can be heard when playing 1 second either side of the click, it sometimes helps to change the playback speed to determine if the click heard is before or after the cursor location which can help you to reposition the cursor toward the click location
• I personally have not been successful in using the spectrum view to locate a click
• Sometimes a click can be heard but I can never locate it – it may actually be something in the music that resembles a click but is not a click.
• On a generally good/clean recordings that exhibit clicks with large dB spikes, I highlight those in yellow and perform auto scan on that limited section to avoid corrections on music nearby that could be corrected as a click.

Identifying clicks is a very challenging problem and I am very happy with the Vinyl Studio program and click repair.  Just stating how difficult it is and no one size fits all, I try different techniques at different times to be successful at finding the clicks.
With all of that said, my question/suggestion is does or can the auto click repair algorithm consider the following when searching for and applying corrections:

• Large change in dB from the adjacent waveform (perhaps a user input in how large the dB change is to be used in scanning)
• Comparison of the waveform to the other channel to help determine if it is similar (maybe not a click) or much different and hence likely a real click.  This may be dependent on how much time either side of the potential click being investigated is considered and possibly a user changeable value.

[Paul Sanders (AlpineSoft)]

Thanks for this.  Believe me, we put a lot of time into fine tuning VinylStudio's click scanner, and amplitude of the waveform is a poor metric for detecting clicks.

As for comparing channels, again, I don't think that would be a reliable indicator that a potential click really is a click.  But thanks for your input.