Fourier Analogue-in-Digital is an archiving-specific codec. It uses DCT by default, with additional solutions to ensure perfect archiving.
FrAD is a codec that nobody owns, nobody patents, and is standardised in so much detail that anyone can easily implement it in any language, even if the original implementation is lost. It also supports tags, cover art, and sequential search, and the original implementation can be supported by any OS on any machine that supports Python.
The main difference between any other formats and FrAD is its resilience to corruption and robustness. For many audio files that are branched fron PCM, even a flip of a bit or two can result in popping noise, failure to play the corrupted frame, or even the loss of all subsequent audio information. FrAD, however, provides robust error recovery such that sporadic flipping of a few bytes is not even a concern.
Lossless: Encoding with DCT does not cause any significant difference from the original, and the decoded audio file is equivalent to what was input into the microphone. The DCT profile is not a straightforward use of PCM, so there may be some computational errors, but in the long run it is more resilient to file corruption than PCM. It is also possible to achieve mathematically perfect losslessness by introducing a PCM profile. Additionally, dedicated hardware is envisioned to perform DCT in the analogue domain, which we expect will enable true 'lossless recording'. Each frame includes a 32-bit CRC and optionally ECC to support transmission error detection and recovery.
Fast encoding/decoding: Each frame in Fourier Analogue-in-Digital is all independent. The default profile has no compression process, and the only operation on the signal is a discrete cosine transform, which can maintain decoding speeds of hundreds of times the playback speed, even in a single thread python. This makes playback unobstructed even on low-end hardware, even without dedicated compute units.
'Overly' flexible metadata: Fourier Analogue-in-Digital can include binary data as tags, not just generic tags or cover art. It also uses only one encoding, UTF-8, for tags, so they can be read in any environment.
Search: Fourier Analogue-in-Digital supports per frame and per sample navigation.
Streaming: All frames in Fourier Analogue-in-Digital are independent of the container and independent of each other. No matter how many samples are cut and stitched together, Fourier Analogue-in-Digital can be played back without problems.
Archiving: Fourier Analogue-in-Digital is tremendously resistant to damage. In addition to error detection with CRC, it supports CD-ROM-equivalent or even stronger (Reed-Solomon) ECC, so Fourier Analogue-in-Digital can repair itself even if the file is severely damaged. And even if the stream is corrupted beyond repair via ECC, the discrete cosine transform can protect the original PCM data from damage and produce nearly the same result as before the corruption.
Lossless Compression: Fourier Analogue-in-Digital was developed for uncompressed archiving, and compression is highly vulnerable to damage. There are plenty of good alternatives if preservation is not a concern.(Lossy-compression profile had been developed but is not recommended.)
DRM: The creator of Fourier Analogue-in-Digital, Haמul, values openness and freedom, and DRM is completely antithetical to his ideas. Bluntly speeking, DRM is the equivalent of preparing to turn a consumer's legitimately purchased 'goods' into electron-trash at any moment and make off, and we have no plans to do so, nor will we ever. The application of third-party DRM is also strictly prohibited.