IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
Help. How to evaluate a Mp3's quality? Like noise existed in the f
MaiTiano
post Apr 16 2012, 12:00
Post #1





Group: Members
Posts: 15
Joined: 17-January 11
From: SZ, China
Member No.: 87372



As mentioned in the title.

Now I am evolving into a project. In this project, we have a 128kbps mp3 library and we need to find which files have the inferior quality, such as noise existed, volume is too small etc.

Any advice? It would be better if someone know some open-source project which already has the ability to detect the mp3 files with bad quality.

many thanks.
Go to the top of the page
+Quote Post
halb27
post Apr 16 2012, 12:10
Post #2





Group: Members
Posts: 2414
Joined: 9-October 05
From: Dormagen, Germany
Member No.: 25015



Sounds like you don't have a reference to the original uncompressed music.
If this is true the only thing that can be achieved IMO is looking for issues like temporal smearing with percussion instruments or impulses for electronic music, or other things like unusual tremolo or other effects. But without reference to the original it's often hard to say whether suspected issues are due to mp3 compression or due to an imperfect recording for instance.


--------------------
lame3100m --bCVBR 300
Go to the top of the page
+Quote Post
Nessuno
post Apr 16 2012, 12:25
Post #3





Group: Members
Posts: 422
Joined: 16-December 10
From: Palermo
Member No.: 86562



QUOTE (MaiTiano @ Apr 16 2012, 13:00) *
Now I am evolving into a project. In this project, we have a 128kbps mp3 library and we need to find which files have the inferior quality, such as noise existed, volume is too small etc.


Do you intend lower quality with reference to the lossless source (which, as halb27 already pointed out, is very difficult without the reference), or absolute sound quality in this set, with reference to parameters like, for example, dynamic range, bandwidth, noise etc..., in which case the fact that they are mp3 has less importance?


--------------------
... I live by long distance.
Go to the top of the page
+Quote Post
MaiTiano
post Apr 17 2012, 02:35
Post #4





Group: Members
Posts: 15
Joined: 17-January 11
From: SZ, China
Member No.: 87372



QUOTE (halb27 @ Apr 16 2012, 12:10) *
Sounds like you don't have a reference to the original uncompressed music.
If this is true the only thing that can be achieved IMO is looking for issues like temporal smearing with percussion instruments or impulses for electronic music, or other things like unusual tremolo or other effects. But without reference to the original it's often hard to say whether suspected issues are due to mp3 compression or due to an imperfect recording for instance.



Yep, I do not have the reference to the original lossless music.

My goal is to tell whether a mp3 file has a good quality or not just depends on its file only. No matter the bad quality is caused by mp3 compression or imperfect recording, I just need to tell people that it has good quality or not.
Go to the top of the page
+Quote Post
saratoga
post Apr 17 2012, 02:53
Post #5





Group: Members
Posts: 4723
Joined: 2-September 02
Member No.: 3264



Determining which will sound better to a person is a subjective process, and no one that can be readily done by a machine.

This post has been edited by db1989: Apr 17 2012, 15:20
Reason for edit: removing unnecessary full quote of above post
Go to the top of the page
+Quote Post
.alexander.
post Apr 17 2012, 06:36
Post #6





Group: Members
Posts: 73
Joined: 14-December 06
Member No.: 38681



Robot will never replace man © Ogre
Go to the top of the page
+Quote Post
Nessuno
post Apr 17 2012, 07:42
Post #7





Group: Members
Posts: 422
Joined: 16-December 10
From: Palermo
Member No.: 86562



QUOTE (saratoga @ Apr 17 2012, 03:53) *
Determining which will sound better to a person is a subjective process, and no one that can be readily done by a machine.


So what the "quality" setting of every lossy encoder does actually mean?

This post has been edited by Nessuno: Apr 17 2012, 07:44


--------------------
... I live by long distance.
Go to the top of the page
+Quote Post
MaiTiano
post Apr 17 2012, 08:44
Post #8





Group: Members
Posts: 15
Joined: 17-January 11
From: SZ, China
Member No.: 87372



Yes,I do think that there is a evaluation rule for the music quality. not only the bitrates can be used to evaluate it(actually bitrates can not be viewed as the accurate quality rule neither:) )

This post has been edited by db1989: Apr 17 2012, 15:21
Reason for edit: as in post #5
Go to the top of the page
+Quote Post
Nessuno
post Apr 17 2012, 09:11
Post #9





Group: Members
Posts: 422
Joined: 16-December 10
From: Palermo
Member No.: 86562



QUOTE (MaiTiano @ Apr 17 2012, 09:44) *
Yes,I do think that there is a evaluation rule for the music quality. not only the bitrates can be used to evaluate it(actually bitrates can not be viewed as the accurate quality rule neither:) )


Well, if they are VBR and they have been encoded by the same encoder with the same settings and they all contain similar musical "patterns" (e.g. all symphonic, all piano, all chamber) then actual bitrate might be a very rough estimate of content quality.

Of course "quality" sic et simpliciter is really a very vague criterion....


--------------------
... I live by long distance.
Go to the top of the page
+Quote Post
MaiTiano
post Apr 17 2012, 10:01
Post #10





Group: Members
Posts: 15
Joined: 17-January 11
From: SZ, China
Member No.: 87372



For the simplicity, for the 1st step, at least I want to detect the noise existed in the music which affect the whole listening experience. Any advices?

This post has been edited by db1989: Apr 17 2012, 15:21
Reason for edit: as in post #8
Go to the top of the page
+Quote Post
pdq
post Apr 17 2012, 13:38
Post #11





Group: Members
Posts: 3313
Joined: 1-September 05
From: SE Pennsylvania
Member No.: 24233



QUOTE (Nessuno @ Apr 17 2012, 02:42) *
QUOTE (saratoga @ Apr 17 2012, 03:53) *
Determining which will sound better to a person is a subjective process, and no one that can be readily done by a machine.


So what the "quality" setting of every lossy encoder does actually mean?

In this case quality has a very specific meaning - how close the audible result is to the original.
Go to the top of the page
+Quote Post
IgorC
post Apr 17 2012, 14:28
Post #12





Group: Members
Posts: 1506
Joined: 3-January 05
From: Argentina, Bs As
Member No.: 18803



Starting from previous knowledge will be something reasonable to do. wink.gif
PEAQ.

This post has been edited by IgorC: Apr 17 2012, 14:28
Go to the top of the page
+Quote Post
Nessuno
post Apr 17 2012, 17:38
Post #13





Group: Members
Posts: 422
Joined: 16-December 10
From: Palermo
Member No.: 86562



QUOTE (pdq @ Apr 17 2012, 14:38) *
QUOTE (Nessuno @ Apr 17 2012, 02:42) *

So what the "quality" setting of every lossy encoder does actually mean?

In this case quality has a very specific meaning - how close the audible result is to the original.


Ok, I'm no expert, but it seems to me that "close to the original" it's not just a matter of analitically measuring the distance between two signals: some assumption of what has more impact in the perceived quality of output are also made, like a larger lowpass on input on higher quality settings (so bandwidth is considered a quality index), clipping prevention (so clipping is considered a low quality index), preserve dynamic, apply weaker masking at certain frequencies and so on...

And in general, to develop a perceptual model, some assumption on which signal characteristics have more influence on its perceived quality should be made, shouldn't they?

This post has been edited by Nessuno: Apr 17 2012, 17:40


--------------------
... I live by long distance.
Go to the top of the page
+Quote Post
saratoga
post Apr 17 2012, 17:53
Post #14





Group: Members
Posts: 4723
Joined: 2-September 02
Member No.: 3264



QUOTE (Nessuno @ Apr 17 2012, 12:38) *
And in general, to develop a perceptual model, some assumption on which signal characteristics have more influence on its perceived quality should be made, shouldn't they?


Correct.
Go to the top of the page
+Quote Post
Dynamic
post Apr 18 2012, 11:51
Post #15





Group: Members
Posts: 793
Joined: 17-September 06
Member No.: 35307



QUOTE (MaiTiano @ Apr 16 2012, 11:00) *
Any advice? It would be better if someone know some open-source project which already has the ability to detect the mp3 files with bad quality.


If I wanted a quick-and-dirty survey of quality of MP3 supplied as is, such as a bunch of free releases I had ten years ago, I'd use encSpot (a Windows program that was released freely as encSpot Pro, though I think the source was never opened).

Certain old, bad, encoders such as Xing or Blade can be detected pretty reliably, and will cause the traffic-light indication in encSpot to show red. Other than that, it won't detect the majority of unwise encoder settings.

Newer encoders are a lot better if used wisely. I still hear segments of podcasts with dreadful encoding - usually low bitrate with an excessively high low-pass.

As for the rest, it seems really tough.



--------------------
Dynamic the artist formerly known as DickD
Go to the top of the page
+Quote Post
knutinh
post Apr 18 2012, 11:59
Post #16





Group: Members
Posts: 568
Joined: 1-November 06
Member No.: 37047



QUOTE (MaiTiano @ Apr 17 2012, 11:01) *
For the simplicity, for the 1st step, at least I want to detect the noise existed in the music which affect the whole listening experience. Any advices?

I would claim that I have several recordings with excessive noise that are still subjectively important to me - and examples of low-noise material that is worthless to me.

Sounds to me like trying to detect "good art" automatically. It is going to be hard as long as we dont even have an agreement on what is "good art".

-h
Go to the top of the page
+Quote Post
MaiTiano
post Apr 18 2012, 12:35
Post #17





Group: Members
Posts: 15
Joined: 17-January 11
From: SZ, China
Member No.: 87372



I would like to add a more specific example to illustrate what I want to do here. blink.gif

For example, in the following figure, as the red square pointed, in these area the dB data is not 0 but with a approximate fixed value. Therefore, the music in those red rectangle has bad quality just like 'click' or some other very short pause. These breakpoints make the music sounds very bad, like segmented piece by piece. mad.gif



Let me zoom in and give you a even clear view of those rectangle 'breakpoint'.(Also, as shown in the red rectangle area)

Go to the top of the page
+Quote Post
Dynamic
post Apr 18 2012, 13:22
Post #18





Group: Members
Posts: 793
Joined: 17-September 06
Member No.: 35307



You've defined your problem a bit better.

You might want to look into SoX audio toolkit. It can be run from the commandline or a batch script, making it possible to deal with a whole bunch of files. It's also open source, like Audacity, which you used to plot that waveform.

It has switches to detect silences, which you could modify to make it detect or split that file:
http://sox.sourceforge.net/sox.html
... scroll down to the silence section.

You could either modify the source code to do what you want, or run Sox to decode an appropriate file and see if it produces more than one silence

Your use case is not normal, as few people who care about quality will have such bad files because they'll have chosen secure or database-verified ripping or listened to the output with their ears.

I'm sure you'll need to do some coding yourself.


--------------------
Dynamic the artist formerly known as DickD
Go to the top of the page
+Quote Post
MaiTiano
post Apr 19 2012, 03:27
Post #19





Group: Members
Posts: 15
Joined: 17-January 11
From: SZ, China
Member No.: 87372



QUOTE (Dynamic @ Apr 18 2012, 14:22) *
You might want to look into SoX audio toolkit. It can be run from the commandline or a batch script, making it possible to deal with a whole bunch of files. It's also open source, like Audacity, which you used to plot that waveform.

It has switches to detect silences, which you could modify to make it detect or split that file:
http://sox.sourceforge.net/sox.html
... scroll down to the silence section.

Many thanks, Dynamic.
I have a problem about your advice. As can be seen from the figure, actually, those horizontal wave which I highlighted in the red box is not on the 0 dB line. In other words, the sound track may not be silence in these area, right?
Go to the top of the page
+Quote Post
saratoga
post Apr 19 2012, 03:33
Post #20





Group: Members
Posts: 4723
Joined: 2-September 02
Member No.: 3264



QUOTE (MaiTiano @ Apr 18 2012, 22:27) *
I have a problem about your advice. As can be seen from the figure, actually, those horizontal wave which I highlighted in the red box is not on the 0 dB line. In other words, the sound track may not be silence in these area, right?


Since you can't hear a DC level, that is still silence.
Go to the top of the page
+Quote Post
MaiTiano
post Apr 19 2012, 03:44
Post #21





Group: Members
Posts: 15
Joined: 17-January 11
From: SZ, China
Member No.: 87372



QUOTE (saratoga @ Apr 19 2012, 04:33) *
Since you can't hear a DC level, that is still silence.


Thanks. Actually, I think it is approximate DC tongue.gif
Go to the top of the page
+Quote Post
Dynamic
post Apr 19 2012, 20:41
Post #22





Group: Members
Posts: 793
Joined: 17-September 06
Member No.: 35307



I'd assumed, perhaps wrongly, that you were displaying Waveform (dB) rather than linear, in which case those were probably only a fraction offset from zero, and setting silence detection to -80 dB might even be OK.
The other approach is to plot a spectrogram - which SoX can export as a PNG image. The silences, whether or not they contain a DC offset as shown, would appear as black in all frequencies. That might be easier to interpret or to analyse graphically or numerically, particularly if converted to an ASCII-encoded graphical format like PPM or PGM (part of PNM or Netpbm format family).


--------------------
Dynamic the artist formerly known as DickD
Go to the top of the page
+Quote Post
MaiTiano
post Apr 20 2012, 04:39
Post #23





Group: Members
Posts: 15
Joined: 17-January 11
From: SZ, China
Member No.: 87372



QUOTE (Dynamic @ Apr 19 2012, 21:41) *
I'd assumed, perhaps wrongly, that you were displaying Waveform (dB) rather than linear, in which case those were probably only a fraction offset from zero, and setting silence detection to -80 dB might even be OK.
The other approach is to plot a spectrogram - which SoX can export as a PNG image. The silences, whether or not they contain a DC offset as shown, would appear as black in all frequencies. That might be easier to interpret or to analyse graphically or numerically, particularly if converted to an ASCII-encoded graphical format like PPM or PGM (part of PNM or Netpbm format family).


I would like to give you a even clear figure which includes the Y axis and even zoomed in view in it.



For the advice of using Sox to draw a DC spectrogram, I have three questions:
[1] I can get the spectrogram PNG image by Sox, but is it standing for the spetrum features of the whole song long?
[2] How can I get the correct part of these silence area? For example, 57.25s ~ 57.36s is silence part and it would be more likely to see the all black in all frequencies right within this time period. In other words, if I analysis the waveform between 57.15s~57.36s, maybe the spectrum is not all black in all frequencies, right?

Hope to get your help. Many thanks.
Go to the top of the page
+Quote Post
Dynamic
post Apr 25 2012, 20:44
Post #24





Group: Members
Posts: 793
Joined: 17-September 06
Member No.: 35307



[1] - Yes, I think it's the whole song
[2] - It should be black, except perhaps for the very lowest frequencies, but probably for those also. You can show spectrogram view in Audacity - that should be black during the gaps too.


--------------------
Dynamic the artist formerly known as DickD
Go to the top of the page
+Quote Post
MaiTiano
post May 22 2012, 02:24
Post #25





Group: Members
Posts: 15
Joined: 17-January 11
From: SZ, China
Member No.: 87372



QUOTE (Dynamic @ Apr 25 2012, 21:44) *
[1] - Yes, I think it's the whole song
[2] - It should be black, except perhaps for the very lowest frequencies, but probably for those also. You can show spectrogram view in Audacity - that should be black during the gaps too.


I already have finish the program of detecting the silent part in a music file.
Now, the left work should be distinguishing the silent part in the beginning/end part(as the music file usually has silent part in the beginning/end of it) of music file and the irregular silent part.
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: 24th April 2014 - 19:23