IPB

Welcome Guest ( Log In | Register )

2 Pages V  < 1 2  
Reply to this topicStart new topic
Audio from vinyl (24 / 96-192 khz). Converting/downsampling?
Porcus
post Aug 29 2012, 20:43
Post #26





Group: Members
Posts: 1993
Joined: 30-November 06
Member No.: 38207



But ... if the peak measures to 0.999, and the resampling clips, then the clipping is what? An unwanted artifact of the algorithm?


--------------------
One day in the Year of the Fox came a time remembered well
Go to the top of the page
+Quote Post
pdq
post Aug 29 2012, 21:02
Post #27





Group: Members
Posts: 3450
Joined: 1-September 05
From: SE Pennsylvania
Member No.: 24233



The resampling produced samples that were beyond +-1, but the resampling itself did not produce distortion. It's when you try to represent those values in a container that has inadequate range, and so have to clamp the values to represent them, that the distortion occurs.
Go to the top of the page
+Quote Post
Taishou
post Aug 29 2012, 23:30
Post #28





Group: Members
Posts: 12
Joined: 27-August 12
Member No.: 102679



Well, I decided to leave it as it is.

I tried the original 192 khz file with Audacity and my resampled non "-G" resulting file. Both have 2 or 3 of peaks that touch the "1" on Audacity's scale. And both sound the same. I picked a file that the peaks weren't because of cracks/clicks, but during some high note. Headphones, speakers, whatever, there's no difference. Plus, the peaks are too small to be noticeable. You have to zoom all in until you see milisecs on the scale, and then the peaks last for nothing.

I also tried the song with the -G command in SoX. It makes those peaks sit right just below -1, making a volume reduction of ~0.03-00.5 db. As always, it all sounds the same, no matter how hard I try. And even if I try, the sound from the vinyl, although it's great considering the source, is far from perfect, so I guess those little clippings will never matter.

It must be because of what pdq says, that the original shows a peak of 0.9999 and the resampled file is 1. For example, one file:

Original: -2.77db, 0.999900
Resampled: -2.80db, 1
Resampled with -G: -2.79, 0.999969

Original and resampled look the same in Audacity, the -G one shows a little lower. It's kinda puzzling, there's a 0.03 db difference between original and resampled. Yet the graphs/curves look exactly the same. When using -G the difference is only 0.01 db, but you can see the graph lower (less volume).

In the end everything is too insignificant, and when you see commercial CDs like this



With peaks around 1.10 and -10 db, you feel nothing seems to matter much tongue.gif (yes, I've been reading about loudness wars)
Go to the top of the page
+Quote Post
greynol
post Aug 29 2012, 23:41
Post #29





Group: Super Moderator
Posts: 10338
Joined: 1-April 04
From: San Francisco
Member No.: 13167



Re. compression:
I've seen worse.

At the end of the day it's what you hear that matters. Clipping should be of no concern unless it is actually audible.

This post has been edited by greynol: Aug 29 2012, 23:42


--------------------
Your eyes cannot hear.
Go to the top of the page
+Quote Post
Taishou
post Aug 29 2012, 23:51
Post #30





Group: Members
Posts: 12
Joined: 27-August 12
Member No.: 102679



Indeed. And it's the same on the source, so it's not because resampling (I checked and only the original files with 0.9999 of peak level are the ones SoX reports as clipping). The others with a bit less than that number go without warnings. I think unless you do further editing, it won't matter if it's 1 or 0.9999, it will sound the same.
Go to the top of the page
+Quote Post
greynol
post Aug 30 2012, 02:51
Post #31





Group: Super Moderator
Posts: 10338
Joined: 1-April 04
From: San Francisco
Member No.: 13167



Could it be due to possible inter-sample overs occurring at analog conversion during playback?


--------------------
Your eyes cannot hear.
Go to the top of the page
+Quote Post
Porcus
post Aug 30 2012, 07:35
Post #32





Group: Members
Posts: 1993
Joined: 30-November 06
Member No.: 38207



QUOTE (Taishou @ Aug 30 2012, 00:51) *
it won't matter if it's 1 or 0.9999, it will sound the same.


The 1 is after clipping, and it would have been 1 even if the clipping had been intolerable. Now in this case you have checked that everything is OK (bar a few inaudible milliBels over a few milliseconds), but just don't take 1 as a general sign that all is well.


--------------------
One day in the Year of the Fox came a time remembered well
Go to the top of the page
+Quote Post
krabapple
post Aug 31 2012, 14:08
Post #33





Group: Members
Posts: 2517
Joined: 18-December 03
Member No.: 10538



QUOTE (Woodinville @ Aug 28 2012, 22:27) *
This just goes to make a point I've made in talks every since about 1990: "SNR is Mostly Harmless"



But when it gets too low in online threads, I tend to bail.
Go to the top of the page
+Quote Post
greynol
post Aug 31 2012, 19:48
Post #34





Group: Super Moderator
Posts: 10338
Joined: 1-April 04
From: San Francisco
Member No.: 13167



Off-topic discussion on the SNR or MP3 has been split here:
http://www.hydrogenaudio.org/forums/index....showtopic=96786

Continuation of that discussion here will result in disciplinary action.


--------------------
Your eyes cannot hear.
Go to the top of the page
+Quote Post

2 Pages V  < 1 2
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: 19th December 2014 - 10:28