IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
Criteria to decide audio quality, Ref vs. Test
BJC
post Jul 22 2004, 12:46
Post #1





Group: Members
Posts: 38
Joined: 5-May 04
Member No.: 13890



R.wav - reference audio PCM
T.wav - test audio PCM
Any criteria or formula to show the quality of T.wav refer to R.wav ?
Go to the top of the page
+Quote Post
dev0
post Jul 22 2004, 12:54
Post #2





Group: Developer
Posts: 1679
Joined: 23-December 01
From: Germany
Member No.: 731



Are you looking for a subjective or objective form of evaluation?

Read up about ABX tests.
You may also want to take a look at EAQUAL.


--------------------
"To understand me, you'll have to swallow a world." Or maybe your words.
Go to the top of the page
+Quote Post
BJC
post Jul 22 2004, 13:04
Post #3





Group: Members
Posts: 38
Joined: 5-May 04
Member No.: 13890



QUOTE
It is not proof of quality, but it is proof of equal or disparate quality.

ABX seems not to tell how good the T.wav is.
Especially while there are two test files, e.g. T1.wav and T2.wav.
Go to the top of the page
+Quote Post
dev0
post Jul 22 2004, 13:05
Post #4





Group: Developer
Posts: 1679
Joined: 23-December 01
From: Germany
Member No.: 731



What you are looking for is an ABC/HR testing tool like ff123's.


--------------------
"To understand me, you'll have to swallow a world." Or maybe your words.
Go to the top of the page
+Quote Post
smack
post Jul 22 2004, 13:15
Post #5





Group: Members
Posts: 182
Joined: 16-January 02
Member No.: 1046



huh.gif It depends on what you mean with "quality".

One method could be a plain file compare:
diff (or fc.exe) T.wav R.wav
If the files are equal then T.wav has 100% quality, otherwise it is crap. blink.gif

More sophisticated methods could be:
-subtract T.wav from R.wav (sample-by-sample) and calculate the power of the difference.wav
-perform listening (blind) tests to find out if there are any audible differences (see other posts for links)
Go to the top of the page
+Quote Post
BJC
post Jul 23 2004, 01:35
Post #6





Group: Members
Posts: 38
Joined: 5-May 04
Member No.: 13890



AFAIK, some high quality audio codecs always use bit-lost/bit-precision to show how good quality they are.
I wonder it's the only method for measuring audio quality. unsure.gif
Is there anything else? If yes, what is the methodology?
Go to the top of the page
+Quote Post
ChangFest
post Jul 23 2004, 07:48
Post #7





Group: Members
Posts: 423
Joined: 3-February 04
Member No.: 11743



That depends on what you mean by quality. Better quality could mean less encoded errors in a file vs another or it could be subjective quality where you believe something sounds better than something else but in reality it could have more errors than the file it's being tested against.
Go to the top of the page
+Quote Post
BJC
post Jul 23 2004, 17:37
Post #8





Group: Members
Posts: 38
Joined: 5-May 04
Member No.: 13890



QUOTE (ChangFest @ Jul 23 2004, 02:48 PM)
... or it could be subjective quality where you believe something sounds better than something else but in reality it could have more errors than the file it's being tested against.
*


So subjective and objective tests are both required to judge the quality of audio files ?
Or a objective test dominates the quality? huh.gif
Go to the top of the page
+Quote Post
Gecko
post Jul 23 2004, 18:07
Post #9





Group: Members
Posts: 934
Joined: 15-December 01
From: Germany
Member No.: 662



Please describe more precisely what it is you wish to do.

If you want to compare the quality of an mp3 encode to the original then there is no exact mathematical method. EAQUAL tries to do this but is far from perfect.

EAQUAL is objective in the sense that with the same input both you and I will obtain the exact same results whereas our opinions might differ strongly once we actually listen to the sample.

In terms of audio, quality can be thought of as "comfort". You could give an exact objective description of a chair but one of us might find the chair more comfortable than the other. The problem is that objective and subjective don't neccessarily correlate and in our case the subjective impression is the most important one.
Go to the top of the page
+Quote Post
BJC
post Jul 24 2004, 18:33
Post #10





Group: Members
Posts: 38
Joined: 5-May 04
Member No.: 13890



AFAIK, some audio decoders have the following requirements to their decoders' implementation.
1. absolute value of the samples, ref(n) and test(n), couldn't excess N. (bit precision)
2. maximum value of the noise(difference) to signal ratio couldn't greater than E (db).

I doubt evev one implements a decoder with bit loss M, for example, equals N+3.
The listeners have the ability to tell which is better than another ?

So i would like to know if there are another methods out of there?
Go to the top of the page
+Quote Post
Hias
post Jul 26 2004, 22:35
Post #11





Group: Members
Posts: 7
Joined: 13-July 04
Member No.: 15398



What do you want to compare?

Bit precision is used as a measure for fixed-point implementations of e.g. decoders - compared to a reference implementation.

Or do you need a perceptual measure?

Then there's e.g. http://www.opticom.de who seem to have quality analyzers based on perceptual models.
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: 25th April 2014 - 04:08