The 'T' in "FFT" stands for Transform. This patent is not about the transform, but about the encoding of the frequency data. FFT libraries, whether open source or closed, do not encode the raw frequency data. FFT libraries store frequency domain data in an array of complex numbers, without any meta data or other encoding.
This would be equivalent to the difference between RAW audio sample files with no headers, versus AIFF, RIFF/WAV, CAF, or other encodings of the audio samples. It's far easier to share audio using standard file format encodings. The patent even cites graphics file formats as an example. Raw pixel data is difficult to process without standard formats to encode the data and metadata. Caveat: The language of patents is often distinct from the terms used in engineering. So, it's possible that I misinterpreted this patent. Brian Willoughby On Sep 1, 2021, at 14:39, Zhiguang Zhang <[email protected]> wrote: > > Hey music-dsp list > > > Any idea why someone would want to patent essentially what an open source FFT > library (and I can name several) does? > > https://patents.google.com/patent/US11024322B2 > > Disclaimer: this US patent is from a previous employer > > > -ez >
