IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
Verify that hires lossless files aren't just upsampled CDA
Anakunda
post Jan 27 2014, 20:44
Post #1





Group: Members
Posts: 460
Joined: 24-November 08
Member No.: 63072



HI!!
I need some advice how to distinguish if 24-bit lossless files aren't just upsampled CD-Audio.
Recommended tool or guide for studying spectral graphs would be highly appreciated.
Go to the top of the page
+Quote Post
DVDdoug
post Jan 27 2014, 21:20
Post #2





Group: Members
Posts: 2588
Joined: 24-August 07
From: Silicon Valley
Member No.: 46454



Since you can't tell by listening, what does it matter?

There are some clues, but there is no way to know for sure, and it can be faked.

In theory, I should be easy to check if the bottom 8-bits contain zeros, but I don't know of a tool that checks that. If there are zeros in the bottom 8-bits, you can be sure it was upsampled. (Or, it could have been recorded in 24-bits from using a 16-bit ADC.)

QUOTE
I need some advice how to distinguish if 24-bit lossless files aren't just upsampled CD-Audio.
If you have FLAC or ALAC, a "fake" 24-bit file should compress to (about) the same size as a compressed CD. (~6-7 MB per minute).

QUOTE
Recommended tool or guide for studying spectral graphs would be highly appreciated.
The spectrum (frequency content) is NOT related to bit depth. It's related to the sample rate.

The simplest explanation is that in order to represent a "wave", you need at least one sample-point for the top-half of the wave and one sample-point for the bottom half of the wave. That means 44.1kHz CD audio simply cannot have audio greater than 22,050Hz. If you have audio above 22,050Hz, and you are sure it's not a measurement or conversion artifact, then your true sample rate is above 44.1kHz.

However, just because there is no audio above 22,050Hz, that does NOT mean it's been upsampled. There are plenty of places in the recording & production chain where ultrasonics may be filtered-out. (And it's a good idea to do so, because you generally shouldn't waste bits, bandwidth, or headroom on stuff you can't hear.)

It would be easy to "fake" the bit depth... Reducing the volume by 0.1dB would probably introduce enough rounding error to fill-in the bottom 8-bits with data. Converting to MP3 and back (with a 24-bit decoder) should accomplish the same thing.

Or if you upsample both the sample rate and the bit depth, (say from 16/44.1 to 24/96) you will fill the 8 bottom bits with interpolated values and you'll never know.

It's a bit trickier, but it's also possible to add harmonics and/or noise to "fake" a higher original sample rate. i.e. Some people claim you need high sample rates to encode vinyl, but you are just encoding ultrasonic noise & distrotion that "looks good" on the spectrum .

This post has been edited by DVDdoug: Jan 27 2014, 21:43
Go to the top of the page
+Quote Post
Anakunda
post Jan 27 2014, 21:46
Post #3





Group: Members
Posts: 460
Joined: 24-November 08
Member No.: 63072



QUOTE (DVDdoug @ Jan 27 2014, 21:20) *
If you have audio above 22,050Hz, and you are sure it's not a measurement or conversion artifact, then your true sample rate is above 44.1kHz.

Many thanks for exhaustive answer! This explains alot for me. The frequency range goes up to 24kHz at 48kHz audio which should be ok.
About the bit depth, I'm afraid that I'm not able to see if bottom 8 bits are filled with zeros, but assuming that sample rate is true should be enough for bit depth too.
Go to the top of the page
+Quote Post
saratoga
post Jan 27 2014, 21:59
Post #4





Group: Members
Posts: 4959
Joined: 2-September 02
Member No.: 3264



Usually up sampling will introduce noise that will fill the lower 8 bits anyway, so checking them often doesn't tell much.
Go to the top of the page
+Quote Post
dhromed
post Jan 28 2014, 10:29
Post #5





Group: Members
Posts: 1314
Joined: 16-February 08
From: NL
Member No.: 51347



QUOTE (DVDdoug @ Jan 27 2014, 21:20) *
Since you can't tell by listening, what does it matter?


Money.
Go to the top of the page
+Quote Post
2Bdecided
post Jan 28 2014, 10:43
Post #6


ReplayGain developer


Group: Developer
Posts: 5134
Joined: 5-November 01
From: Yorkshire, UK
Member No.: 409



While there are several tell-tale clues which would suggest or even prove that the source was CD quality audio, the absence of those clues does not prove that the source was native 24-bit or native (any specific sample rate). It's easy enough to hide all these clues, accidentally or on purpose.

That said, I've read (not verified myself) that some "hi rez" releases (24/96) are quite obviously upsampled, due to a hard cut off in the spectrogram at 22kHz. 16-bit dither during digital fade outs, and/or 16-bit noise shaping, would be detectable too, even if the bottom 8-bits were no longer zeros due to some other processing.

Cheers,
David.
Go to the top of the page
+Quote Post
bennetng
post Jan 28 2014, 19:06
Post #7





Group: Members
Posts: 224
Joined: 22-December 05
Member No.: 26587



What I worry about is some releases may make use of DSP effects to artifically add lower bit information and high frequency harmonics to fake customers if the sellers know how others determine a release's authenticity.

This post has been edited by bennetng: Jan 28 2014, 19:18
Go to the top of the page
+Quote Post
Porcus
post Jan 28 2014, 19:32
Post #8





Group: Members
Posts: 1842
Joined: 30-November 06
Member No.: 38207



Some remastering engineers would likely even believe that letting some algorithm add high frequency harmonies, will make it closer to what was performed in the studio or concert hall, by making a sufficiently clever guess of what the "missing" parts would sound like ...

(The might even be right, and in a way that could even have mattered if it had happened in the audible range.)


--------------------
One day in the Year of the Fox came a time remembered well
Go to the top of the page
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 17th September 2014 - 08:00