Audio/Video Codec Quality verify tool.

[Boki]

So many codecs,

For audio, whatever codec, finally, it should be decode to wav form to
output, right?

When codec based data be converted to wav, we can check its different
with original wav data, right?

For Video, it's the same.

Does that not a way to verify difference ?


Best regards,
Boki.

 

[martin griffith]

So many codecs,

For audio, whatever codec, finally, it should be decode to wav form to
output, right?

When codec based data be converted to wav, we can check its different
with original wav data, right?

For Video, it's the same.

Does that not a way to verify difference ?


Best regards,
Boki.

Yes, You can subtract the audio that has been processed from the
original and listen. This will give a general idea of the codec
quality. It is very difficult to put a value on the signal
degradation in numeric form that means anything (IMHO)

Try snell and wilcox web site for video information


martin

 

[Pooh Bear]

So many codecs,

For audio, whatever codec, finally, it should be decode to wav form to
output, right?

When codec based data be converted to wav, we can check its different
with original wav data, right?

For Video, it's the same.

Does that not a way to verify difference ?

It'll give the mathematical difference for sure. It would certainly be
interesting to listen to ! I'm unsure how effectively it would be in
rating the psycho-acoustic effect of different codecs though.

Graham

 

[dalai lamah]

Un bel giorno Pooh Bear digitò:

I'm unsure how effectively it would be in
rating the psycho-acoustic effect of different codecs though.

I would say "near zero". It has been shown that either time- and
frequency-domain comparisons are ineffective. Double blind listening tests
seem to be the correct way. This is one of the advocates of this technique:

http://ff123.net/

 

[Boki]

Thanks a lot for information.

Best regards,
Boki.

 

[Boki]

If we don't care the human ear avalible range ( 20~20K), it is a
mathematical result.
umm. .. I want to say,

If we calculate that, we can know whcih codec has less lost, isn't it ?

Best regards,
Boki.

 

[Pooh Bear]

If we don't care the human ear avalible range ( 20~20K), it is a
mathematical result.
umm. .. I want to say,

If we calculate that, we can know whcih codec has less lost, isn't it ?

In a mathematical sense - yes.

That may not correlate well to the effect perceived by human hearing
though.

Your idea is very interesting though. If you choose to develop it, I would
be very interested to hear the results.


Graham

 

[Boki]

haha... in fact, I think that is a time problem, but the more important
I want to care is the final result is useful or not.

About 2~3 years before, I was developing Video/Image processing, I
think the numerical difference didn't get too much sense for human
eyes, certainly it is a reference.

I think we don't have to pay effort on the data processing, we have so
many sound tool in our world already, just read them ( decoded
mp3/wmv/...) from our program to calculate what we want.

It seems that something is related to ART

Best regards,
Boki.