Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Some 2L Hi-Res music samples are either botched or cooked (Read 3221 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Some 2L Hi-Res music samples are either botched or cooked

I'm not sure if it is due to sloppy work or intentional fudging, but for the second time now in just a few days I've noticed that the songs I happened to download from 2L's HiRes Download - test bench samples page, to compare with Foobar ABX testing, have exhibited audible flaws, tells, which have nothing to do with sampling rates, bit depth, or MQA deblurring.

Today's song, the 16bit/44k vs 24bit/96k versions of Ola Gjeilo's North Country II, has a minor time synch error which is easily detectable using Foobar ABX test's phrase repeat function:
Code: [Select]
foo_abx 2.0.1 report
foobar2000 v1.3.9
2016-03-14 14:30:50

File A: 2L-048_14_stereo_44kHz_16b.flac
SHA1: 007639ed4d83823e1e72f6029c3f52f3f7af8f05
File B: 2L48SACD_14_stereo_96k.flac
SHA1: 4171ca9fd626c42eef0e1d91deabce8dd66acdbc

Output:
DS : Primary Sound Driver
Crossfading: NO

14:30:50 : Test started.
14:33:59 : 01/01
14:34:08 : 02/02
14:34:19 : 03/03
14:34:26 : 04/04
14:34:31 : 05/05
14:34:36 : 06/06
14:34:42 : 07/07
14:34:52 : 08/08
14:35:00 : 09/09
14:35:06 : 10/10
14:35:06 : Test finished.

 ----------
Total: 10/10
Probability that you were guessing: 0.1%

 -- signature --
345f879a467e817ec7b36e0ad5614b3f05857bf0
Note, it is just a few milliseconds out of time alignment so a naïve listener might not even consciously pick up on it with slow, casual switching, but that's exactly how to best cook a demo: fool them into thinking what they are keying on is the "Hi-Re$" part since they can't quite put their finger on the distinction unless you point it out to them. [Minor level changes on the order of a fraction of a dB are notorious for creating the exact same illusion: you hear a subtle change but you don't perceive it as level. Just like with the McGurk effect, knowledge of the illusion doesn't make you immune to it either, even if you are one of the world's foremost experts on the topic].

If the very first two songs I happened to download to compare both show flaws I'd be rather surprised if there aren't at least some others as well. So the next time some newb brags that they can hear a difference on this site's files, send them on over to this thread.

Re: Some 2L Hi-Res music samples are either botched or cooked

Reply #1
When I A/B-ed a DSD-256 file from that site (down-sampled to 32/192 due to DAC limitations) against both the 16/44.1 version AND a 16/44.1 version of the DSD file mentioned above created by myself, the differences were fairly pronounced.
What one thinks they hear, perceptually, and what their ear actually hears are two different things.
Quote
How do you analyze files for time alignment errors?
In the case of the particular 2L song in the above post I can simply hear the time synch error when I switch back and forth, but in the case of the more subtle time alignment error I heard in the AIX records' song Mosaic A2 vs B2, which I discussed in detail at the AVS forums' thread on the challenge (I go by m. zillch over there), I first heard the error but couldn't quite put my finger on what I was hearing. So I then measured it with Audacity and then posted a video showing an example of how it was manifested on a particular guitar riff, making it easier for others to comprehend:
Mosaic Time Error video

This image shows the left tracks only from Mosaic A2 and B2, with the 10 - 11ms or so time misalignment: