Digitizing a continuous waveform requires measuring it many times per second; that count is the sample rate. The complete Nyquist–Shannon condition is narrower than the common slogan: the signal must be band-limited, and the sample rate must be greater than twice its highest frequency for ideal reconstruction. Real recording systems also require an anti-aliasing filter, so “2×” alone is not an unconditional quality guarantee.
If content above the Nyquist frequency is not removed first, it aliases into lower frequencies and cannot be separated from the sampled result afterwards.
Covering frequencies up to roughly 20 kHz requires a rate above 40 kHz plus room for an anti-aliasing filter transition band. The 44.1 kHz rate became widespread through CD audio; its exact value is also tied to early engineering systems that recorded digital audio through video infrastructure.
The MP3 standard, ISO/IEC 11172-3, supports three sample rates — 32, 44.1, and 48 kHz — and 44.1 kHz remains the mainstream for music distribution.
The MP3 Cutter processes the PCM produced by the browser decoder. MP3 export performs another lossy encode. WAV export avoids adding another lossy encode, but it cannot restore information already discarded by the source format.
Try the MP3 Cutter NowdecodeAudioData() resamples to the AudioContext sample rate when required).