[Industrial One]

How do I turn http://i38.tinypic.com/33xu8p1.png into http://i34.tinypic.com/2zftt13.png

?

And what did the perpetrator do to the waveform? It sounds exactly the
same as the proceeding length. But If I amplify, then some parts of
the weird-looking 3-second waveform in the beginning are inaudible.

[[email protected]]

On Aug 8, 12:18 pm, Industrial One wrote:
How do I turnhttp://i38.tinypic.com/33xu8p1.pngintohttp://i34.tinypic.com/2zftt13.png

And what did the perpetrator do to the waveform? It sounds exactly the
same as the proceeding length. But If I amplify, then some parts of
the weird-looking 3-second waveform in the beginning are inaudible.

The first example clearly showws the first segment
has a significant DC offset. The waveform also exhibits
some slipped samples

The second example shows that DC offset removed, and
the clipped samples remain.

To turn the first into the second, using any of the common
editors about (e.g., Audacity, select that portion of the wave,
and (usually) under the "Normalize" menu function you'll
find an option to "Remove DC offsett."

The more interesting question, if you're interested,
is how did it get thatw ay to begin with.

The most likely scenario is someone may have started
with a 3-second audio file recorded in offset binary
format (the default for 8-bit linear PCM wave files)
and then appended a two'-s complement segment to
that.

An offset binary file takes has no "sign": It takes a
normal signed two's complement representation
(that, say, in 8-bit, goes from -128 to +127), adds
128 to it (so the result is no 0 to 255).

It's possible that whatever editor was used to
append these segment failed to renormalize the
offset binary segment properly.

[Industrial One]

On Aug 8, 11:04 am, wrote:
On Aug 8, 12:18 pm, Industrial One wrote:

How do I turnhttp://i38.tinypic.com/33xu8p1.pngintohttp://i34.tinypic.com/2zftt13.png

And what did the perpetrator do to the waveform? It sounds exactly the
same as the proceeding length. But If I amplify, then some parts of
the weird-looking 3-second waveform in the beginning are inaudible.

The first example clearly showws the first segment
has a significant DC offset. The waveform also exhibits
some slipped samples

Clipped, you mean?

The second example shows that DC offset removed, and
the clipped samples remain.

To turn the first into the second, using any of the common
editors about (e.g., Audacity, select that portion of the wave,
and (usually) under the "Normalize" menu function you'll
find an option to "Remove DC offsett."

Kickass, this fixed it. Thanks.

The more interesting question, if you're interested,
is how did it get thatw ay to begin with.

The most likely scenario is someone may have started
with a 3-second audio file recorded in offset binary
format (the default for 8-bit linear PCM wave files)
and then appended a two'-s complement segment to
that.

An offset binary file takes has no "sign": It takes a
normal signed two's complement representation
(that, say, in 8-bit, goes from -128 to +127), adds
128 to it (so the result is no 0 to 255).

It's possible that whatever editor was used to
append these segment failed to renormalize the
offset binary segment properly.

You're saying that that 3-second sample is 8-bit while the rest of the
audio is 16-bit?

[[email protected]]

On Aug 8, 1:39 pm, Industrial One wrote:
On Aug 8, 11:04 am, wrote:
It's possible that whatever editor was used to
append these segment failed to renormalize the
offset binary segment properly.

You're saying that that 3-second sample is 8-bit while the rest of the
audio is 16-bit?

No, I'm saying that's one plausible way it could have
gotten that way.

[Industrial One]

On Aug 8, 12:00 pm, wrote:
On Aug 8, 1:39 pm, Industrial One wrote:

On Aug 8, 11:04 am, wrote:
It's possible that whatever editor was used to
append these segment failed to renormalize the
offset binary segment properly.

You're saying that that 3-second sample is 8-bit while the rest of the
audio is 16-bit?

No, I'm saying that's one plausible way it could have
gotten that way.

I think it was done on purpose, 'cuz that 3-second part is where the
DJ's name was whispered, so he wanted to make it harder to remove --
or something. Problem is, that part has a really high amplitude which
forces the rest of the audio to be really low (btw, how can it have a
higher amplitude yet have the same amount of decibels as the rest of
the waveform?)

[[email protected]]

On Aug 8, 3:12 pm, Industrial One wrote:
I think it was done on purpose, 'cuz that 3-second part is where the
DJ's name was whispered, so he wanted to make it harder to remove --
or something. Problem is, that part has a really high amplitude which
forces the rest of the audio to be really low (btw, how can it have a
higher amplitude yet have the same amount of decibels as the rest of
the waveform?)

Because if the thing that measures decibels essentially
has a high-ass filter function to it, the DC offset is
irrelevant.

[Richard Crowley]

"Industrial One" wrote ...
I think it was done on purpose, 'cuz that 3-second part is where the
DJ's name was whispered, so he wanted to make it harder to remove --

Seems much more likely that somebody tried to boost the signal
level (to hear the faint audio), but didn't do a very good job of it.