opus VS AppleNeuralHash2ONNX

Opus doesn't support 44.1 kHz because compatibility and effort/benefit ratio:

https://github.com/xiph/opus/issues/43

The browser audio limitation is presumably a workaround to some bug or performance limitation that was relevant at some point in history (the site was created in 2014).

There is a pattern here (that also goes with the author's prior article on inverting gauss' sum formula): Generally if if you can make a formula that counts the combination of things you can convert that into a code to encode and decode those combinations into indexes.

So for example the opus audio codec needs to encode/decode vectors of dimension n whos absolute values sum to k. https://github.com/xiph/opus/blob/master/celt/cwrs.c#L74

Or this rolling cuckoo filter that optimally encode/decode four sorted numbers in a range 0..2N with the constraint that the they span a range of N. https://github.com/sipa/bitcoin/blob/202006_cuckoo_filter/sr...

If you're lucky there will be closed form expressions for the encoding and decoding equations. (There for both of the above, at least for some parameters, but in both those examples the implementations use small tables because for the ranges involved the tables end up being faster than sqrts).

Too bad 10Hz is a too slow to generate audio-rate bitops music.

(e.g. https://github.com/xiph/opus/blob/master/tests/test_opus_enc... )

Could anyone help me get a disk image files for older Global Underground CDs? I encoded my old CDs into subpar mp3 files, and I'd now like to have high-quality Opus encodings and experiment across various bitrates.

Presumably, OP is referring to the Opus audio codec versus Dolby's AC3 codec.

Indeed. https://opus-codec.org/

If there are multiple tags with the same name, Ffmpeg will only use the last tag. If you really need to have multiple tags with the same name in your OPUS files, use opusenc instead (https://opus-codec.org/). Beware that some playback software does not display multiple artists gracefully.

AFAIK ogg isn't really suitable for low latency audio streaming. Consider the Opus codec instead.

I have asked on GitHub also https://github.com/xiph/opus/issues/263 in anyone wants to respond there.

Perhaps you're referring to the hash collisions that were generated for the client-side NeuralHash which was reverse engineered about a year ago: https://github.com/AsuharietYgvar/AppleNeuralHash2ONNX/issues/1

Except Apple’s hashes have collisions where images that are similar in content but not identical will have the same hash. Meaning their process will have the same result: benign images being flagged as CSAM. See an example here https://github.com/AsuharietYgvar/AppleNeuralHash2ONNX/issues/1

perceptual hashes are completely broken due to false positives, like making a picture of a dog having the same hash as CSAM images. They are vulnerable to collision and have no preimage resistance, the most important feature. Because of that you can DOS conversations and people by sending dog images with the same hash as a CSAM: https://github.com/AsuharietYgvar/AppleNeuralHash2ONNX/issues/1

Specifically the fact that you can have one hash on two images, for example. There are other ways to abuse the CSAM system, which is confirmed by several researchers (one, two). And Apple saying it's not a concern... only to pull back on these plans.