Piano Forum

One way of avoiding clipping is to record using 32 bit floating point, making use of its headroom.  It is rather useful in live recordings where there might not be an opportunity to do retakes.  That said, it is not an excuse to avoid doing due diligence and checking audio levels beforehand. 

That seems a little harsh... there's a learning curve to doing your own recording and I found that out the hard way, but I guess it is important, as not doing it can wreck a very good recording session.

I was referring to recording using 32 bit floating point, in which it is not an unrecoverable disaster to have peaks higher than 0dB.  There is less worry about peaks above 0dB as one can recover by simply attenuating the wav, the data is still there you just have to adjust the file.  In comparison, fixed point bitrates one cannot record data higher than 0dB. 

To put things  in perspective here are some dynamic ranges you could expect from common bit rates:
16 bit fixed point: 96.3 dB
24 bit fixed point: 144.5 dB
32 bit floating point: 1528 dB

For 32 bit floating point the noise floor is -758 dB and the max value is 770 dB. 

Speaking pragmatically, 32 bit floating point provides a range far beyond what would be reasonably required to capture musical sounds within a digital file.  It allows a more flexible post production workflow because you can adjust the file in post, to either boost quiet sounds or attenuate ones that go over 0 dB.  It is much more forgiving in live recordings when you have the possibility of unexpected peaks, or unknown sound levels from sources. 

That said, good recording technique is always good practice.  One should not become lazy just because the technology allows it.  With fixed point bit rates, one should absolutely check levels before recording, it is an essential requirement. 

Placing the dynamic range of recorded sounds correctly was one of the first things I learned when starting recording.  The results of not doing this can be disastrous and frustrating to the music being recorded.  Out of experience, I quickly learned the importance of understanding this.

***

Back to the OP, I am leaning towards the problem not being clips as a result of gain being set too high.  If you listen to the first few seconds, the attack of the opening chord does not produce a pop, yet there are small pops during the much quieter sustain portion of the chord.  There are bigger pops during the single note melody, again much softer in volume than the opening chord, very unlikely caused by clips due to high gain.