How to tell "true" wave files

How to tell "true" wave files

Is there any software to tell whether an audio track was actually
converted from an MP3 or WMA file, rather than ripped from a CD?

Re: How to tell "true" wave files

In article <c46b389d-d52e-418b-969c-3548a509b226@z24g2000prf.googlegroups . com >,
Dennis Q. Wilson <DennisQWilson@yahoo . com > wrote:
>Is there any software to tell whether an audio track was actually
>converted from an MP3 or WMA file, rather than ripped from a CD?

If your ears won't tell you, looking at a spectrogram probably will. If you
see big blocky squares... that's a sign of compression.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Re: How to tell "true" wave files

> In article <c46b389d-d52e-418b-969c-3548a509b226@z24g2000prf.googlegroups . com >,
> Dennis Q. Wilson <DennisQWilson@yahoo . com > wrote:
> >Is there any software to tell whether an audio track was actually
> >converted from an MP3 or WMA file, rather than ripped from a CD?
>
<snip>
There are several software apps that purport to do exactly that.
"Trader's Little Helper" is one. I'm guessing that they work by looking
for the same characteristics mentioned elsewhere in this thread. This is
a big deal in the lossless live music trading community, where polluting
the dataspace with lossy compressed sources is strictly verboten.

Peace
Art

Re: How to tell "true" wave files

"Dennis Q. Wilson" <DennisQWilson@yahoo . com > wrote in
message
news:c46b389d-d52e-418b-969c-3548a509b226@z24g2000prf.googlegroups . com

> Is there any software to tell whether an audio track was
> actually converted from an MP3 or WMA file, rather than
> ripped from a CD?

AFAIK, detecting a MP3 is a study in audio forensics. You collect evidence
and infer a conclusion from it.

Two kinds of evidence that are fairly reliable:

(1) Spectrogram shows a brick-wall roll-off, often at 16 KHz

(2) MP3 frames are usually 417 or 418 bytes, so a wav file that is based on
an MP3 should be an integer number of frames long.

If the bitrate was low enough, the poor fidelity is more evidence.

Re: How to tell "true" wave files

"Dennis Q. Wilson" <DennisQWilson@yahoo . com > wrote in message
news:c46b389d-d52e-418b-969c-3548a509b226@z24g2000prf.googlegroups . com ...
> Is there any software to tell whether an audio track was actually
> converted from an MP3 or WMA file, rather than ripped from a CD?

Probably the easiest way is to use a program which has a spectrum analyzer.
(Adobe Audition is one, there are many others.) If there's a sharp cutoff at
around 15kHz then the file came from a lowish-bit-rate .mp3 or .wma file. If
the original was high-bitrate, though, it's harder to tell.

Peace,
Paul

Re: How to tell "true" wave files

On Tue, 6 May 2008 11:24:32 -0700 (PDT), "Dennis Q. Wilson"
<DennisQWilson@yahoo . com > wrote:

>Is there any software to tell whether an audio track was actually
>converted from an MP3 or WMA file, rather than ripped from a CD?

Have a look at this:

* 81.174.169.10/odds/identmp3.gif

It is the spectrogram of a few seconds of music on a CD, followed by
the same piece following 256k conversion to MP3. The differences are
obvious, the main one being the complete lack of low level detail
above 16kHz - only the peaks are showing. Apart from that, the whole
appearance of the MP3 section is more granular which is evidence of
the amount of data you can discard because it is masked. Audibly, the
two pieces are essentially identical.

The piece I have used is opera with virtually no amplitude compression
- that being the kind of stuff that MP3 does best; it really doesn't
like having to cope with hypercompressed pop music that doesn't
contain the kind of low level detail that can be masked away.

Anyway, what this shows is that visually it is very easy to tell a
file that has been through MP3 conversion from one that hasn't.

d

--
Pearce Consulting
* w w w .pearce.uk . com

Re: How to tell "true" wave files

On May 6, 4:23 pm, nos...@nospam . com (Don Pearce) wrote:
> On Tue, 6 May 2008 11:24:32 -0700 (PDT), "Dennis Q. Wilson"
>
> <DennisQWil...@yahoo . com > wrote:
> >Is there any software to tell whether an audio track was actually
> >converted from an MP3 or WMA file, rather than ripped from a CD?
>
> Have a look at this:
>
> * 81.174.169.10/odds/identmp3.gif
>
> It is the spectrogram of a few seconds of music on a CD, followed by
> the same piece following 256k conversion to MP3. The differences are
> obvious, the main one being the complete lack of low level detail
> above 16kHz - only the peaks are showing. Apart from that, the whole
> appearance of the MP3 section is more granular which is evidence of
> the amount of data you can discard because it is masked. Audibly, the
> two pieces are essentially identical.
>
> The piece I have used is opera with virtually no amplitude compression
> - that being the kind of stuff that MP3 does best; it really doesn't
> like having to cope with hypercompressed pop music that doesn't
> contain the kind of low level detail that can be masked away.
>
> Anyway, what this shows is that visually it is very easy to tell a
> file that has been through MP3 conversion from one that hasn't.
>
> d
>
> --
> Pearce Consulting * w w w .pearce.uk . com

hey look at that!

the MP3 got rid of all the hiss :-)

Mark

Re: How to tell "true" wave files

"Mark" <makolber@yahoo . com > wrote in message
news:a16a362d-649f-41c4-92fd-b43296cb0fea@z72g2000hsb.googlegroups . com
> On May 6, 4:23 pm, nos...@nospam . com (Don Pearce) wrote:
>> On Tue, 6 May 2008 11:24:32 -0700 (PDT), "Dennis Q.
>> Wilson"
>>
>> <DennisQWil...@yahoo . com > wrote:
>>> Is there any software to tell whether an audio track
>>> was actually converted from an MP3 or WMA file, rather
>>> than ripped from a CD?
>>
>> Have a look at this:
>>
>> * 81.174.169.10/odds/identmp3.gif
>>
>> It is the spectrogram of a few seconds of music on a CD,
>> followed by the same piece following 256k conversion to
>> MP3. The differences are obvious, the main one being the
>> complete lack of low level detail above 16kHz - only the
>> peaks are showing.

Agreed and expected.


>> Apart from that, the whole appearance
>> of the MP3 section is more granular which is evidence of
>> the amount of data you can discard because it is masked.

I don't see that at all. What I see is the usual granular nature of a
spectrogram with the parameters used.

>> Audibly, the two pieces are essentially identical.

>> The piece I have used is opera with virtually no
>> amplitude compression - that being the kind of stuff
>> that MP3 does best; it really doesn't like having to
>> cope with hypercompressed pop music that doesn't contain
>> the kind of low level detail that can be masked away.

>> Anyway, what this shows is that visually it is very easy
>> to tell a file that has been through MP3 conversion from
>> one that hasn't.

> hey look at that!

> the MP3 got rid of all the hiss :-)

I wish.

Audible hiss is usually above -60 dB, and down in the 4 KHz range where the
ear is most sensitive.

IME, a 256K MP3 pretty well preserves low level detail down into the -60 dB
range, so its not going to remove audible hiss, and its not going to producw
a spectrogram with appreciable visible changes due to the removal of
low-level detail.

Re: How to tell "true" wave files


> > hey look at that!
> > the MP3 got rid of all the hiss   :-)
>
> I wish.
>
> Audible hiss is usually above -60 dB, and down in the 4 KHz range where th=
e
> ear is most sensitive.
>
> IME, a 256K MP3 pretty well preserves low level detail down into the -60 d=
B
> range, so its not going to remove audible hiss, and its not going to produ=
cw
> a spectrogram  with appreciable visible changes due to the removal of
> low-level detail.-

Arnie... I fully agree, I was just trying to be funny...

I have tried A/B comparisons several times and I have not ever heard
any difference between an original and a 256K MP3.

At 128K they can get swishy...

Mark

Re: How to tell "true" wave files

Mark <makolber@yahoo . com > wrote:
>
>I have tried A/B comparisons several times and I have not ever heard
>any difference between an original and a 256K MP3.

Listen for stereo imaging changes. If you have a good speaker system that
actually images properly, it is very interesting to hear what the encoder
does to the stereo field. Stuff starts moving around.

>At 128K they can get swishy...

Yes, that's when it starts to get _really_ nasty.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Re: How to tell "true" wave files

Scott Dorsey wrote:
> Mark <makolber@yahoo . com > wrote:
>>
>> I have tried A/B comparisons several times and I have not ever heard
>> any difference between an original and a 256K MP3.
>
> Listen for stereo imaging changes. If you have a good speaker system
> that actually images properly, it is very interesting to hear what
> the encoder does to the stereo field. Stuff starts moving around.
>
>> At 128K they can get swishy...
>
> Yes, that's when it starts to get _really_ nasty.


And that's what most of them are....

geoff

Re: How to tell "true" wave files

Dennis Q. Wilson schreef:
> Is there any software to tell whether an audio track was actually
> converted from an MP3 or WMA file, rather than ripped from a CD?

To add to the others' good advice, here is one trick. In the audio
application of your choice, open the track, then flip the left and right
channels, then invert the phase. Then play this together with the
original track. If the original track has been compressed to a low bit
rate somewhere along the way, you will hear artifacts all over the place.

HTH

Jos.

--
Ardis Park Music
w w w .ardispark.nl

Re: How to tell "true" wave files

On May 6, 2:37 pm, Jos Geluk <skelt...@casema.invalid> wrote:
> Dennis Q. Wilson schreef:
>
> > Is there any software to tell whether an audio track was actually
> > converted from an MP3 or WMA file, rather than ripped from a CD?
>
> To add to the others' good advice, here is one trick. In the audio
> application of your choice, open the track, then flip the left and right
> channels, then invert the phase. Then play this together with the
> original track. If the original track has been compressed to a low bit
> rate somewhere along the way, you will hear artifacts all over the place.
>
> HTH
>
> Jos.
>
> --
> Ardis Park Musicw w w .ardispark.nl



I discovered that one while knocking out vocals.

The Mp3's have phasing artifacts all over the place and the vocals
don't disappear.

Re: How to tell "true" wave files

On May 6, 11:24 am, "Dennis Q. Wilson" <DennisQWil...@yahoo . com >
wrote:
> Is there any software to tell whether an audio track was actually
> converted from an MP3 or WMA file, rather than ripped from a CD?

Thank you, one and all. I have come away from this exchange with a
lot of solid, practical information.

Re: How to tell "true" wave files


Do a search for "AudioChecker v1.2 by Dester"

I use this to ferret out low quality MP3 rips that people recode to FLAC
files. It works reasonably well.


d'bEAR


Quoting "Dennis Q. Wilson" <DennisQWilson@yahoo . com >:

>Is there any software to tell whether an audio track was actually
>converted from an MP3 or WMA file, rather than ripped from a CD?