Help! How do I fix this problem?

[Druhms]

A lady brought in a cassette tape of a deposition wanting me to fix it. She
described the problem to me as her tape was dragging and it needed speeding up
so she could understand the words. Well, I listen to it and it is much worse
than a simple thing like that. It sounds as if her machine went crazy. It
sounds like it has been run through a hardcore chorus with the rate at full
speed. I don't know what caused this. Within one short word it goes from
super slow to super fast about 400 times. Horrible warble sound. Can't make
out anything anyone is saying. Does anyone have any pointers on this? Maybe
some software that can fix this? I can't imagine there being a quick fix.
Seems like the only way to fix it would be to take it word by word andslow down
and speed up each sweep....which would be rediculous. It is very unpredictable
too, so it isn't like you could put on the effect to counter the effect
already there. I hope I'm explaining this well.
Thanks for the help everyone.
JJ

[axtogrind]

This isn't a rec.audio.pro response, but it may help. Usually the most
effecient solution isn't what the client came in looking for. If you tell
her that you estimate it will cost $5,000 for you to go through and recreate
word for word, or whatever you would charge for all that time, you might
discover that she has other sources for the information: for example, is
that the 'original' of the tape? Was there not a reporter / stenographer
present at the time? (Is there a stenographer's tape of the transcript that
was typed at the time.) It is VERY unusual for a deposition to be recorded
on cassette with NO OTHER means of recording. I know some state offices
that do it in my state, of very short (10 - 15 minute) depositions, but that
is it.

'Cause if it is important enough to depose someone, it is important enough
to hire a court reporter.

So if you DON'T get any good tech responses here, you might just have to
give her a minute by minute quote...

Keith

"Druhms" <druhms@aol.com> wrote in message
news:20030825213712.27902.00000605@mb-m29.aol.com...
> A lady brought in a cassette tape of a deposition wanting me to fix it.
She
> described the problem to me as her tape was dragging and it needed
speeding up
> so she could understand the words. Well, I listen to it and it is much
worse
> than a simple thing like that. It sounds as if her machine went crazy.
It
> sounds like it has been run through a hardcore chorus with the rate at
full
> speed. I don't know what caused this. Within one short word it goes from
> super slow to super fast about 400 times. Horrible warble sound. Can't
make
> out anything anyone is saying. Does anyone have any pointers on this?
Maybe
> some software that can fix this? I can't imagine there being a quick fix.
> Seems like the only way to fix it would be to take it word by word andslow
down
> and speed up each sweep....which would be rediculous. It is very
unpredictable
> too, so it isn't like you could put on the effect to counter the effect
> already there. I hope I'm explaining this well.
> Thanks for the help everyone.
> JJ

[Jerry Steiger]

"Druhms" <druhms@aol.com> wrote in message
news:20030825213712.27902.00000605@mb-m29.aol.com...
> A lady brought in a cassette tape of a deposition wanting me to fix it.
She
> described the problem to me as her tape was dragging and it needed
speeding up
> so she could understand the words. Well, I listen to it and it is much
worse
> than a simple thing like that. It sounds as if her machine went crazy.
It
> sounds like it has been run through a hardcore chorus with the rate at
full
> speed. I don't know what caused this. Within one short word it goes from
> super slow to super fast about 400 times. Horrible warble sound. Can't
make
> out anything anyone is saying. Does anyone have any pointers on this?
Maybe
> some software that can fix this?

I don't know squat about this, but I seem to remember a discussion here or
in the Pro Audio newslist of a possible solution. If you assume the bias
frequency was relatively constant, then you could use it to control the
playback speed. How you actually do this is left as an exercise for the
reader, as this mechanical engineer doesn't have a clue.

Jerry Steiger

[W. Williams]

I've seen this happen with really cheap tape recorders when they start to
malfunction. The occasions I have seen this tape actually speeds up at
random and when recording and then slows down to normal speed, although your
example appears to be more extreme.

I have successfully restored short selections (up to 20 minutes), using the
pitch bend function in Sound Forge, to something approaching useable,
although artefacts are unavoidable. Most WAV editing software will have a
similar function. This is a pretty time-consuming exercise and should only
be undertaken if it is fairly short or it is a really important and your
client is prepared to pay for a less than ideal final product. You could
offer to try a short section of say a minute, see if you are able to obtain
acceptable results, and then extrapolate a quote for the entire file.

You basically have to get the best possible recording you can to your PC for
editing in your software of choice (16/44 should be fine). The method I use
is to work in small sections starting from the beginning of the file. Look
for the first "normal" spot and then select the area up to that point - I
like to create regions, but markers will work just as well. Open your pitch
bend function (usually found under the effects menu) and then manually add,
preview and adjust envelope points until your selection sounds acceptable.
Then you update the file and repeat the process.

My experience was that generally the audio would rapidly increase in pitch
till it reached a specific point, where it would remain constant for a few
seconds before decreasing in the same manner. This results in the envelope
having a more or less \___/ shape. This will obviously be inverted if you
have to increase the pitch.

I hope my understanding of your problem is correct and, if so, that you have
some success.

W

"Druhms" wrote:

> A lady brought in a cassette tape of a deposition wanting me to fix it.
She
> described the problem to me as her tape was dragging and it needed
speeding up
> so she could understand the words. Well, I listen to it and it is much
worse
> than a simple thing like that. It sounds as if her machine went crazy.
It
> sounds like it has been run through a hardcore chorus with the rate at
full
> speed. I don't know what caused this. Within one short word it goes from
> super slow to super fast about 400 times. Horrible warble sound. Can't
make
> out anything anyone is saying. Does anyone have any pointers on this?
Maybe
> some software that can fix this? I can't imagine there being a quick fix.
> Seems like the only way to fix it would be to take it word by word andslow
down
> and speed up each sweep....which would be rediculous. It is very
unpredictable
> too, so it isn't like you could put on the effect to counter the effect
> already there. I hope I'm explaining this well.
> Thanks for the help everyone.
> JJ

[Scott Dorsey]

Druhms <druhms@aol.com> wrote:
>A lady brought in a cassette tape of a deposition wanting me to fix it. She
>described the problem to me as her tape was dragging and it needed speeding up
>so she could understand the words. Well, I listen to it and it is much worse
>than a simple thing like that. It sounds as if her machine went crazy. It
>sounds like it has been run through a hardcore chorus with the rate at full
>speed. I don't know what caused this. Within one short word it goes from
>super slow to super fast about 400 times. Horrible warble sound. Can't make
>out anything anyone is saying. Does anyone have any pointers on this? Maybe
>some software that can fix this? I can't imagine there being a quick fix.
>Seems like the only way to fix it would be to take it word by word andslow down
>and speed up each sweep....which would be rediculous. It is very unpredictable
>too, so it isn't like you could put on the effect to counter the effect
>already there. I hope I'm explaining this well.
>Thanks for the help everyone.

So, it's rapid warble? Is it all in recording, or is it also in playback?
If you watch the tape path, is anything bouncing around? Does she have the
original machine that it was made with so you can take it apart and see what
was originally going on?
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[P Stamler]

And to add to Scott's question, have you transplanted the tape to a new
cassette with new slip-sheets and pressure pad, and did that make any
difference?

Peace,
Paul

[Ricky W. Hunt]

"W. Williams" <W.Williams@HotPOP.com> wrote in message
news:1061896210.668915@skink.ru.ac.za...

> Open your pitch bend function (usually found under the effects menu) and
then manually add,
> preview and adjust envelope points until your selection sounds acceptable.
> Then you update the file and repeat the process.

If it's happening 400 times a word (which I'm pretty sure is an extreme
rustication) that wouldn't even if you have the patience to do it.

[Druhms]

Thanks everyone for your responses, and to answer Scott's questions, it is a
very fast warble. I'm assuming it is in playback. It was recorded onto the
tape like this by her machine. I admit I am exagerating when I say 400 times
per word, but more realistically about 80 times per word. Very fast. I would
think it would be very painful going through and fixing 30 minutes of this
with a pitch control device in Pro Tools or Sound Forge. I was hoping there
was some restoration plug in or some trick that a forensics person may use.
Thanks again,
JJ



>So, it's rapid warble? Is it all in recording, or is it also in playback?
>If you watch the tape path, is anything bouncing around? Does she have the
>original machine that it was made with so you can take it apart and see what
>was originally going on?
>--scott
>
>--
>"C'est un Nagra. C'est suisse, et tres, tres precis."

[Scott Dorsey]

Druhms <druhms@aol.com> wrote:
>Thanks everyone for your responses, and to answer Scott's questions, it is a
>very fast warble. I'm assuming it is in playback. It was recorded onto the
>tape like this by her machine.

Is it in playback or is it only in recording? If it's also in playback,
you can fix that. But if it's only on the recording, you can't do anything
about that.

>I admit I am exagerating when I say 400 times
>per word, but more realistically about 80 times per word. Very fast. I would
>think it would be very painful going through and fixing 30 minutes of this
>with a pitch control device in Pro Tools or Sound Forge. I was hoping there
>was some restoration plug in or some trick that a forensics person may use.

No, but there might be a way to fix it if it's a mechanical issue that is
only on playback, and if it's a mechanical issue on recording you can at
least see what is going on to see about undoing it. Does the pressure pad
bounce when you play it back?
--scott
>


--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Mike Rivers]

In article <20030826164755.02751.00000367@mb-m23.aol.com> druhms@aol.com writes:

> Thanks everyone for your responses, and to answer Scott's questions, it is a
> very fast warble. I'm assuming it is in playback. It was recorded onto the
> tape like this by her machine.

Sounds like the recorder was momentarily slowing down, practically
stopping. When played back at a reasonably constant speed, it would
sound like a rapid rise in pitch.

> but more realistically about 80 times per word. Very fast. I would
> think it would be very painful going through and fixing 30 minutes of this
> with a pitch control device in Pro Tools or Sound Forge.

So what's wrong with a little pain if you're going to be well paid for
it? This sort of thing doesn't get done automagically. And forensics
restorers get very well paid. Try a bit and see how much time it takes
you to make a couple of sentences listenable, then give your client a
quote for the job. All she can do is say "no thanks."




--
I'm really Mike Rivers - (mrivers@d-and-d.com)

[dmills@spamblock.demon.co.uk]

Druhms <druhms@aol.com> wrote:
> A lady brought in a cassette tape of a deposition wanting me to fix it. She
> described the problem to me as her tape was dragging and it needed speeding
> up so she could understand the words. Well, I listen to it and it is much
> worse than a simple thing like that. It sounds as if her machine went
> crazy. It sounds like it has been run through a hardcore chorus with the
> rate at full speed. I don't know what caused this. Within one short word
> it goes from super slow to super fast about 400 times. Horrible warble
> sound. Can't make out anything anyone is saying. Does anyone have any
> pointers on this? Maybe some software that can fix this? I can't imagine
> there being a quick fix. Seems like the only way to fix it would be to
> take it word by word andslow down and speed up each sweep....which would
> be rediculous. It is very unpredictable too, so it isn't like you could
> put on the effect to counter the effect already there. I hope I'm
> explaining this well.

I would start by transfering the tape to a new shell to ensure it is not
a friction problem on playback, then if that doesn't work it gets
interesting...

Digitise the signal directly off the tape playback head with a FAST ADC,
(you want the capture the bias signal), the lock a software PLL with
a reasonably wide lockup range and a loop bandwidth sufficient to track
the speed variations to the bias signal, use the control loop signal to
control your playout speed. You could also of course do this in the
hardware domain with electronics built on a bit of veroboard and
with the control loop acting as a servo loop controlling the motor
speed..
Removing the flywheel on the playback deck will make the loop
stability problem more manageable.

A fair bit of work in C/Fortran/Matlab or whatever your weapon of choice
is for this sort of computer play but it MAY be doable....

I have done similar things with old 1/2 inch reel to reel video
(Tape baking wouldn't fix it, and the speed was very variable, fixed by
measuring the actual time between line sync pulses then resampling the
line to fit).

Regards, Dan.
--
** The email address *IS* valid, do NOT remove the spamblock
And on the evening of the first day the lord said...........
..... LX 1, GO!; and there was light.

[Mike Rivers]

In article <u82mib.ms.ln@spamblock.demon.co.uk> dmills@spamblock.demon.co.uk writes:

> I would start by transfering the tape to a new shell to ensure it is not
> a friction problem on playback, then if that doesn't work it gets
> interesting...

Too late for that. Since the description of the problem was that it
gets "super fast" on playback, unless there's a serious problem in the
playback deck, the damage was done on recording - the tape slowed down
when it was recorded, so that when played back at the correct speed,
it speeds up.

I don't think there's any automatic way to fix this problem. That's
why this sort of work is very expensive (or it's done by people who
regret taking on the job).



--
I'm really Mike Rivers - (mrivers@d-and-d.com)

[Mike Rivers]

In article <tkotib.7i2.ln@spamblock.demon.co.uk> dmills@spamblock.demon.co.uk writes:

> I still think you could use the bias to control a servo loop.

I suppose you could, but then you'd have to build a transport that can
respond to speed changes that quickly. You'd probably have better luck
with something like generating a "speed" track, and then writing a
pitch shift algorithm that takes its imput from the speed.

Some FM instrumentation recorders use the carrier frequency to derive
a flutter compensation sigal. Flutter, when the FM is demodulated,
manifests itself as amplitude noise.


--
I'm really Mike Rivers - (mrivers@d-and-d.com)