Last post by zoomorph -
2.17: * Ignore/skip releases which 404 when updating tags. * Wait longer with each successive http 429. * Fix a problem with arrays only getting depth 1 when they should have depth 2, causing certain tagging strings not to work as they should. * Fix wrongly parsing single track mix releases which use "1.x" for every track position. * Skip "(silence)" when parsing hidden tracks, but add the duration of silence to the track duration. * Add the duration of hidden tracks to the track duration. This is returned by DISCOGS_DURATION_SECONDS and used in matching tracks to files by duration.
Given the data provided by the above cited references we can conclude that phase distortion is indeed audible, though generally speaking, only very subtly so and only under certain specific test conditions and perception circumstances.
Impressive as the paper(?) looks like, I'm not sure about how qualified they are to draw that conclusion.
Last post by kode54 -
I had to reset something with SELinux that was preventing the forums from invoking the sendmail process, which randomly popped up on a software upgrade. I officially hate SELinux even more than when I had to set up the exemptions just to get the forum script working in the first place.
If this happens again, I am outright disabling SELinux. Piece of crap.
Last post by Phanton_13 -
Also there is the possibility of bad caps (old electrolytic ones that dried as example) in the filters chain doing weird things with the signal. Transistor, ICs, diodes and resistors can also go bad and affect signal but its much less frequent at least in the audible range of frequencies.
I came across a dude who thinks he can hear phase distortions made by an equalizer. Putting aside differences in loudness of different frequencies, this means he should be able to hear differences made by an all-pass filter. I doubt he will prove it even if he could. But I am pretty sure this difference is impossible to hear if tested correctly (so, for example, signal after filtering doesn't clip, etc). I tried to search the web a bit and could not find if anyone succeeded in detecting an all-pass filter by ear in a blind ABX test. Should I immediately dismiss this dude as yet another placebo troll, or he might actually have golden ears?
I don't know about any particular EQ, but it's definitely possibly to hear phase distortion if it's sufficiently large. Taking an extreme example, imagine a system that has zero phase from 0 to 2 kHz, followed by a steep slope after that. It would basically mean that the HF get delayed compared to the LF. If you make that delay large enough (e.g. 100 ms), you're definitely going to hear it. I'm not sure where that threshold is. Similarly, if you have a stereo signal and you have different phase distortion to the two channels in the LF, then again you're going to hear it as a spatial shift in the stereo image. The sensitivity to phase depends a lot on the signal. If your signal is just two constant tones, then I don't think you can hear anything at all. OTOH, if your signal is made of clear/narrow impulses, then you want good phase response to avoid temporal smearing. It's all about whether the phase distortion changes the temporal shape of the signal.
I have thought a little about putting in 64-bit support (I assume float64 only), but it's really low in my priority and I will be too busy to do anything that significant for the foreseeable future, so I kinda doubt it. There are some architectural issues (like that fact that WavPack currently processes audio in 32-bit wide buffers), and there's the issue that real 64-bit float audio data (in other words, audio that's been processed so that all the mantissa bits are used, as opposed to data that's really just something smaller stored in a 64-bit float) is not going to compress very well at all. So, if you can typically get 32% compression on 32-bit floats, I would expect only about 16% on real 64-bit floats.
Of course, one of the nice things about WavPack float compression is if it detects that, for instance, the audio is really only 16-bit, then it compresses it to the equivalent amount, and this would be even more useful if you were storing in 64-bit floats.
If this becomes popular and people start clamoring for it, I would certainly reconsider, but so far this is the first clamor.
Hey! Thanks I won't even be using 64fp myself, was just a test and until there is a real advantage to using it I'll be sticking with 32... But just thought it may benefit others and make wavpack a bit more future proof.... Certainly no need for it, thanks again for your amazing compression!!! I have about 11tb of multitrack wav projects archived away, I'd need a second Nas box if it wasn't for wavpack!