Sign in with
Sign up | Sign in
Your question

Stretching WAV files without losing quality

Last response: in Home Audio
Share
Anonymous
October 12, 2004 8:30:45 AM

Archived from groups: rec.audio.pro,rec.video.desktop (More info?)

I have a 2 hour WAV file that I need to make 1 second longer. I was doing
the video capture and audio was ahead of video by 1 second factor.

So I need to stretch it without losing quality. The effect should be the
same as if I played an analog cassette tape a bit slower. The reason I'm asking
is that I had never had problems with squeezing WAV file (i.e. making them
shorter). But I once tried to stretch the WAV file in GoldWave sound editor
and artifacts such as clicks and distortion were introduced in sound. Any way
to do it without artifacts? Thanks.

--Leonid

PS. The WAV file is 16 bit 48kHz.
Anonymous
October 12, 2004 8:30:46 AM

Archived from groups: rec.audio.pro,rec.video.desktop (More info?)

Leonid Makarovsky wrote:

> I have a 2 hour WAV file that I need to make 1 second longer. I was doing
> the video capture and audio was ahead of video by 1 second factor.
>
> So I need to stretch it without losing quality. The effect should be the
> same as if I played an analog cassette tape a bit slower. The reason I'm asking
> is that I had never had problems with squeezing WAV file (i.e. making them
> shorter). But I once tried to stretch the WAV file in GoldWave sound editor
> and artifacts such as clicks and distortion were introduced in sound. Any way
> to do it without artifacts? Thanks.

Time/pitch scaling has really matured in the last couple of
years years. Steinberg's Wavelab has one of the best
implementations I've encountered. Celemony is a product
made specifically for that purpose, has a lot more
capability than you probably need, is rather expensive, but
has a hell of a reputation. Adobe Audition in its latest
incarnation is also reputed to be quite good but is outside
my experience because it requires XP.


Bob
--

"Things should be described as simply as possible, but no
simpler."

A. Einstein
Anonymous
October 12, 2004 10:09:26 AM

Archived from groups: rec.audio.pro,rec.video.desktop (More info?)

Leonid Makarovsky wrote:
> I have a 2 hour WAV file that I need to make 1 second longer. I was
> doing
> the video capture and audio was ahead of video by 1 second factor.
>
> So I need to stretch it without losing quality. The effect should be
> the
> same as if I played an analog cassette tape a bit slower. The reason
> I'm asking is that I had never had problems with squeezing WAV file
> (i.e. making them shorter). But I once tried to stretch the WAV file
> in GoldWave sound editor and artifacts such as clicks and distortion
> were introduced in sound. Any way to do it without artifacts? Thanks.
>
> --Leonid
>
> PS. The WAV file is 16 bit 48kHz.


Sound Forge will do this very easily - and without the clicks and pops. For
that matter, the audio tools built into Vegas will do it as well and you'd
have the added benefit of seeing what you're doing. You can get a trial
version of either to see if they do what you want.

Mike
Related resources
Anonymous
October 12, 2004 12:03:18 PM

Archived from groups: rec.audio.pro,rec.video.desktop (More info?)

Cooledit, Wavelab Soundforge etc will all do a respectanle job, although you
might consider putting in snippets of silence if you can

Don

"Leonid Makarovsky" <venom@cs.bu.edu> wrote in message
news:ckfmll$qi4$1@news3.bu.edu...
> I have a 2 hour WAV file that I need to make 1 second longer. I was doing
> the video capture and audio was ahead of video by 1 second factor.
>
> So I need to stretch it without losing quality. The effect should be the
> same as if I played an analog cassette tape a bit slower. The reason I'm
asking
> is that I had never had problems with squeezing WAV file (i.e. making them
> shorter). But I once tried to stretch the WAV file in GoldWave sound
editor
> and artifacts such as clicks and distortion were introduced in sound. Any
way
> to do it without artifacts? Thanks.
>
> --Leonid
>
> PS. The WAV file is 16 bit 48kHz.
Anonymous
October 12, 2004 1:59:18 PM

Archived from groups: rec.audio.pro (More info?)

Hi Leonid,

> I have a 2 hour WAV file that I need to make 1 second longer. I was doing
> the video capture and audio was ahead of video by 1 second factor.
> So I need to stretch it without losing quality. The effect should be the
> same as if I played an analog cassette tape a bit slower. The reason I'm asking
> is that I had never had problems with squeezing WAV file (i.e. making them
> shorter). But I once tried to stretch the WAV file in GoldWave sound editor
> and artifacts such as clicks and distortion were introduced in sound. Any way
> to do it without artifacts? Thanks.

I never used GoldWave but 1 second on a 7200 second WAVE file means you
need a minimal stretch to 100.013% of the original. Any time stretching
algorithm should be able to handle that without artifacts - especially
if you don't need to preserve pitch. It might be that GoldWave is
struggling because of the size of the file, but may be someone else can
comment on that.

Cheers,
Walco
Anonymous
October 12, 2004 3:39:39 PM

Archived from groups: rec.audio.pro,rec.video.desktop (More info?)

On Tue, 12 Oct 2004 01:34:59 -0700, Bob Cain
<arcane@arcanemethods.com> wrote:

>
>
>Leonid Makarovsky wrote:
>
>> I have a 2 hour WAV file that I need to make 1 second longer. I was doing
>> the video capture and audio was ahead of video by 1 second factor.
>>
>> So I need to stretch it without losing quality. The effect should be the
>> same as if I played an analog cassette tape a bit slower. The reason I'm asking
>> is that I had never had problems with squeezing WAV file (i.e. making them
>> shorter). But I once tried to stretch the WAV file in GoldWave sound editor
>> and artifacts such as clicks and distortion were introduced in sound. Any way
>> to do it without artifacts? Thanks.
>
>Time/pitch scaling has really matured in the last couple of
>years years. Steinberg's Wavelab has one of the best
>implementations I've encountered. Celemony is a product
>made specifically for that purpose, has a lot more
>capability than you probably need, is rather expensive, but
>has a hell of a reputation. Adobe Audition in its latest
>incarnation is also reputed to be quite good but is outside
>my experience because it requires XP.

Magix, who market Samplitude, sell a less powerful version called
"Magix Studio" or something like that for around $50, it's a very
powerful app that misses a few of the high end features of Samplitude,
but the time stretch should be in there and it works very well. I
think you can download a trial version and check it out. Also there
is a good time-stretch plugin from Prosoniq called "Time Factory" that
is excellent.

Al
Anonymous
October 12, 2004 4:01:42 PM

Archived from groups: rec.audio.pro,rec.video.desktop (More info?)

In article <ckfmll$qi4$1@news3.bu.edu> venom@cs.bu.edu writes:

> I have a 2 hour WAV file that I need to make 1 second longer. I was doing
> the video capture and audio was ahead of video by 1 second factor.

Is this one second (cumulative) off at the end of the two hour video
program? Why not just find a few places where the sync is far enough
to be noticable (like a sound that doesn't occur right in time with
the image of what makes the sound), chop the audio file, and drag it
into place? Line up all the important parts and the rest of it won't
matter. If you have to make the time come out exact, you can always
find other places to chop it. This is easier, less potentially
damaging to the audio (but back up your file first before cutting it)
and will give you a better production.


--
I'm really Mike Rivers (mrivers@d-and-d.com)
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me here: double-m-eleven-double-zero at yahoo
Anonymous
October 12, 2004 6:25:07 PM

Archived from groups: rec.audio.pro,rec.video.desktop (More info?)

In rec.audio.pro Mike Kujbida <kujfam-misleadingspam@sympatico.ca> wrote:
: Sound Forge will do this very easily - and without the clicks and pops. For
: that matter, the audio tools built into Vegas will do it as well and you'd
: have the added benefit of seeing what you're doing. You can get a trial
: version of either to see if they do what you want.

Mike,

I have SoundForge 5.0. What option of stretching should I use? There're many
there. I also have the tool called SSRC which is the best tool for resampling.
I was thinking maybe I should resample this WAV file to say 48.001kHz and then
tell the program to play it back at 48kHz?

--Leonid
Anonymous
October 12, 2004 6:26:00 PM

Archived from groups: rec.audio.pro,rec.video.desktop (More info?)

In rec.audio.pro Don Nafe <dnafe@magma.ca> wrote:
: Cooledit, Wavelab Soundforge etc will all do a respectanle job, although you
: might consider putting in snippets of silence if you can

What does that mean?

--Leonid
Anonymous
October 12, 2004 6:26:01 PM

Archived from groups: rec.audio.pro,rec.video.desktop (More info?)

"Leonid Makarovsky" <venom@cs.bu.edu> wrote in message
news:ckgpho$hoa$2@news3.bu.edu...
> In rec.audio.pro Don Nafe <dnafe@magma.ca> wrote:
> : Cooledit, Wavelab Soundforge etc will all do a respectanle job, although
> you
> : might consider putting in snippets of silence if you can
>
> What does that mean?

Cut the audio track apart at several natural pauses and slip it back
into sync so that the error doesn't accumulate.
Anonymous
October 12, 2004 7:08:47 PM

Archived from groups: rec.audio.pro (More info?)

On Tue, 12 Oct 2004 07:35:03 -0700, "Richard Crowley"
<rcrowley7@xprt.net> wrote:

>
>"Leonid Makarovsky" <venom@cs.bu.edu> wrote in message
>news:ckgpho$hoa$2@news3.bu.edu...
>> In rec.audio.pro Don Nafe <dnafe@magma.ca> wrote:
>> : Cooledit, Wavelab Soundforge etc will all do a respectanle job, although
>> you
>> : might consider putting in snippets of silence if you can
>>
>> What does that mean?
>
>Cut the audio track apart at several natural pauses and slip it back
>into sync so that the error doesn't accumulate.
>
And fill those stretch marks with "room sound", the background noise
you recorded when nobody was speaking.

Mike T.
Anonymous
October 12, 2004 7:14:34 PM

Archived from groups: rec.audio.pro (More info?)

Mike T. <miket@invalid.net> wrote:
:>Cut the audio track apart at several natural pauses and slip it back
:>into sync so that the error doesn't accumulate.
:>
: And fill those stretch marks with "room sound", the background noise
: you recorded when nobody was speaking.

To be precisely correct it's not even a second I have to stretch this file more
but .155 of a second. Sounds like a small thing, however, for the synchronizing
audio and video it's 5 frames of NTSC video and it will be noticeable. So
really the best thing for me is just to make audio slower to match it.

--Leonid
Anonymous
October 12, 2004 7:14:35 PM

Archived from groups: rec.audio.pro (More info?)

In article <ckgscq$84n$1@news3.bu.edu>,
Leonid Makarovsky <venom@cs.bu.edu> wrote:
>Mike T. <miket@invalid.net> wrote:
>:>Cut the audio track apart at several natural pauses and slip it back
>:>into sync so that the error doesn't accumulate.
>:>
>: And fill those stretch marks with "room sound", the background noise
>: you recorded when nobody was speaking.
>
>To be precisely correct it's not even a second I have to stretch this file more
>but .155 of a second. Sounds like a small thing, however, for the synchronizing
>audio and video it's 5 frames of NTSC video and it will be noticeable. So
>really the best thing for me is just to make audio slower to match it.

That's more like 1/6 second to my mind, but it's close. Rather than
slowing down the audio, you can remove little chunks here and there between
words. If you do it evenly and cleanly, it can match up nicely.

It's much easier in the video world than it was with film. With 35mm film
you can only remove 1/4 frame since you have to line the perfs up. This is
1/96 second, which can sometimes be an awful lot in a musical piece although
it is reasonable enough for cutting dialogue. 16mm was even worse since you
had to take out one frame at a time from the sound mag since there was only
one set of perfs per frame.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Anonymous
October 12, 2004 7:57:29 PM

Archived from groups: rec.audio.pro (More info?)

> To be precisely correct it's not even a second I have to stretch this file
> more
> but .155 of a second. Sounds like a small thing, however, for the
> synchronizing
> audio and video it's 5 frames of NTSC video and it will be noticeable. So
> really the best thing for me is just to make audio slower to match it.

Seems to me that resampling the entire piece could introduce all sorts of
new sampling errors, where inserting a few extra spaces doesn't require all
that...it's just a bit shift in time.

Another thought...did the sync go bad proportionally or at some specific
point?

-John O
Anonymous
October 12, 2004 8:31:37 PM

Archived from groups: rec.audio.pro,rec.video.desktop (More info?)

Leonid Makarovsky wrote:

> I have SoundForge 5.0. What option of stretching should I use? There're many
> there. I also have the tool called SSRC which is the best tool for resampling.
> I was thinking maybe I should resample this WAV file to say 48.001kHz and then
> tell the program to play it back at 48kHz?

SRC will give an aritfact free result but the the frequency
content of the material is also scaled. If you want the
pitch to remain unchaged, SRC is not the way to do it.

Time/pitch scaling is much different than source rate
conversion (although SRC may be part of the internal
process). SRC has a theoretical solution that can be
calculated to any desired accuracy. Time/pitch scaling
doesn't have a theoretical solution (it's non-causal and
non-linear in really nasty ways) and is approached instead
with various hueristics. The heuristics have dramatically
improved in recent years for almost all products that
include the capability.


Bob
--

"Things should be described as simply as possible, but no
simpler."

A. Einstein
Anonymous
October 12, 2004 8:54:05 PM

Archived from groups: rec.audio.pro (More info?)

Leonid Makarovsky wrote:

> Mike T. <miket@invalid.net> wrote:
> :>Cut the audio track apart at several natural pauses and slip it back
> :>into sync so that the error doesn't accumulate.
> :>
> : And fill those stretch marks with "room sound", the background noise
> : you recorded when nobody was speaking.
>
> To be precisely correct it's not even a second I have to stretch this file more
> but .155 of a second. Sounds like a small thing, however, for the synchronizing
> audio and video it's 5 frames of NTSC video and it will be noticeable. So
> really the best thing for me is just to make audio slower to match it.

In that case, the very slight pitch change that SRC will
introduce probably won't be of any signifigance to you so
should work even better than time/pitch scaling. If, that
is, you can find an SRC app that allows you to specify the
change as the ratio of two time durations rather than just
as a target sample rate. That could require a bunch of
zeros after the decimal point before you start getting
numbers and I've found few programs that use all the digits
it allows you to type in. Madening that.


Bob
--

"Things should be described as simply as possible, but no
simpler."

A. Einstein
Anonymous
October 12, 2004 10:39:59 PM

Archived from groups: rec.audio.pro (More info?)

John O <johno@#no^spam&heathkit.com> wrote:
: Seems to me that resampling the entire piece could introduce all sorts of
: new sampling errors, where inserting a few extra spaces doesn't require all
: that...it's just a bit shift in time.

: Another thought...did the sync go bad proportionally or at some specific
: point?

The sync goes bad proportionally due to capture card vs soundcard internal
clock differences. Capture card captures video frames based on its internal clock
and maybe speed of the VCR whereas soundcard just samples analog sound
independently.

But I just found the solution I'm going to try based on this article:
http://www.videohelp.com/forum/userguides/140540.php

In SoundForge I will SET the sample rate (will not do the actual resampling)
to 47,998 (or something like that) to create a new length and save the new WAV
file. Then I will resample with SSRC.exe program back to 48kHz. I think this
should do the job.

Thanks to everyone who replied.

--Leonid
Anonymous
October 12, 2004 10:40:00 PM

Archived from groups: rec.audio.pro (More info?)

In article <ckh8dv$kj4$1@news3.bu.edu> venom@cs.bu.edu writes:

> The sync goes bad proportionally due to capture card vs soundcard internal
> clock differences.

Awwww, naughty, naughty. Sholda synched 'em. But I suspect that this
might have been a paste-up job.

Still I think that editing the audio is the best way to fix this
rather than time-stretch.


--
I'm really Mike Rivers (mrivers@d-and-d.com)
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me here: double-m-eleven-double-zero at yahoo
Anonymous
October 12, 2004 11:00:53 PM

Archived from groups: rec.audio.pro,rec.video.desktop (More info?)

On 12 Oct 2004 04:30:45 GMT, Leonid Makarovsky <venom@cs.bu.edu>
wrote:

>I have a 2 hour WAV file that I need to make 1 second longer. I was doing
>the video capture and audio was ahead of video by 1 second factor.
>
>So I need to stretch it without losing quality. The effect should be the
>same as if I played an analog cassette tape a bit slower. The reason I'm asking
>is that I had never had problems with squeezing WAV file (i.e. making them
>shorter). But I once tried to stretch the WAV file in GoldWave sound editor
>and artifacts such as clicks and distortion were introduced in sound. Any way
>to do it without artifacts? Thanks.

Playing an analogue tape slower would shift pitch as well as change
length. Though with a change of this tiny magnitude, it's hardly an
issue.

"Audio ahead of video" implies that they just need aligning. But you
mean the overall lengths are different and you really have an audio
file that, though aligned to video at the start, ends up 1 sec.
behind?

1 sec. in two hours. That's a tiny ratio. I'm not sure that any wave
editor will accept a resampling factor of 0.9998611 and, if it did,
do anything useful with it.

Within the 2 hours of audio, can you find 10 places where you could
unobtrusively insert 0.1 sec of silence (or paste in that amount of
ambient sound, if more appropriate? Would that get things near
enough?

CubaseFAQ www.laurencepayne.co.uk/CubaseFAQ.htm
"Possibly the world's least impressive web site": George Perfect
Anonymous
October 12, 2004 11:09:12 PM

Archived from groups: rec.audio.pro,rec.video.desktop (More info?)

Leonid Makarovsky wrote:
> In rec.audio.pro Mike Kujbida <kujfam-misleadingspam@sympatico.ca>
> wrote:
>> Sound Forge will do this very easily - and without the clicks and
>> pops. For that matter, the audio tools built into Vegas will do it
>> as well and you'd have the added benefit of seeing what you're
>> doing. You can get a trial version of either to see if they do what
>> you want.
>
> Mike,
>
> I have SoundForge 5.0. What option of stretching should I use?
> There're many there. I also have the tool called SSRC which is the
> best tool for resampling. I was thinking maybe I should resample this
> WAV file to say 48.001kHz and then tell the program to play it back
> at 48kHz?
>
> --Leonid


Not sure is SF 5.0 has it. I've got SF 6.0 and I'd use the Process > Time
Stretch option myself. I just tried this with a 2 hr. clip and it took
about 2 min. to process. Because the stretch amount is negligible (0.01%),
you won't hear the difference.

Mike
Anonymous
October 13, 2004 2:30:20 AM

Archived from groups: rec.audio.pro (More info?)

> 1 second on a 7200 second WAVE file means you
>need a minimal stretch to 100.013%
>


Audition will reliably handle a number like 100.01389 (7201/7200)
Anonymous
October 13, 2004 4:29:04 AM

Archived from groups: rec.audio.pro,rec.video.desktop (More info?)

In rec.audio.pro Bob Cain <arcane@arcanemethods.com> wrote:
: SRC will give an aritfact free result but the the frequency
: content of the material is also scaled. If you want the
: pitch to remain unchaged, SRC is not the way to do it.


I don't want pitch to remain unchanged. The problem is that SSRC will not
convert a WAV file with 47999Hz to 48000Hz. It says that it must be dividable
by 2 or 3.

So I guess I'm stuck with SoundForge 5.0. I don't like how GoldWave does it.

--Leonid
Anonymous
October 13, 2004 4:30:53 AM

Archived from groups: rec.audio.pro (More info?)

Bob Cain <arcane@arcanemethods.com> wrote:
: In that case, the very slight pitch change that SRC will
: introduce probably won't be of any signifigance to you so
: should work even better than time/pitch scaling. If, that
: is, you can find an SRC app that allows you to specify the
: change as the ratio of two time durations rather than just
: as a target sample rate. That could require a bunch of
: zeros after the decimal point before you start getting
: numbers and I've found few programs that use all the digits
: it allows you to type in. Madening that.

Can you point me to any programms that do that? Or how can SRC do it?
My SSRC version is 1.29.

--Leonid
Anonymous
October 13, 2004 4:30:54 AM

Archived from groups: rec.audio.pro (More info?)

Leonid Makarovsky wrote:

> Can you point me to any programms that do that? Or how can SRC do it?
> My SSRC version is 1.29.

Sorry, I don't know of any SRC programs that allow the kind
of precision specification you need. Sure doen't mean they
don't exist, though.

Is SSRC based on Super Rabbit Code? That is supposed to be
the premier resampling implementation but I'm not sure if it
exists other than as a library routine that apps can call.

You ought to Google it and see what turns up.


Bob
--

"Things should be described as simply as possible, but no
simpler."

A. Einstein
Anonymous
October 13, 2004 4:30:54 AM

Archived from groups: rec.audio.pro (More info?)

Hi Leonid,

<snip>

> Can you point me to any programms that do that? Or how can SRC do it?
> My SSRC version is 1.29.

Audacity (open source, http://audacity.sourceforge.net/) will happily do
this.

Cheers,
Walco
Anonymous
October 13, 2004 4:39:26 AM

Archived from groups: rec.audio.pro (More info?)

Walco <walci@yahoo.com> wrote in
news:416b8ed6$0$36861$e4fe514c@news.xs4all.nl:

> I never used GoldWave but 1 second on a 7200 second WAVE file means
> you need a minimal stretch to 100.013% of the original. Any time
> stretching algorithm should be able to handle that without artifacts -
> especially if you don't need to preserve pitch.

100.013% wil be so close to the original pitch that only a person with
inhumanly perfect pitch will ever hear the difference.

IME, an average person won't notice until about 2% unless comparing side by
side. At 2% the tempo change is as noticeable as the pitch change.

And to answer the OP's question, Adobe Audition (aka Cool Edit) can handle
that comfortably.
Anonymous
October 13, 2004 2:06:45 PM

Archived from groups: rec.audio.pro,rec.video.desktop (More info?)

"Leonid Makarovsky" <venom@cs.bu.edu> skrev i en meddelelse
news:ckfmll$qi4$1@news3.bu.edu...
> I have a 2 hour WAV file that I need to make 1 second longer. I was doing
> the video capture and audio was ahead of video by 1 second factor.
>
> So I need to stretch it without losing quality. The effect should be the
> same as if I played an analog cassette tape a bit slower. The reason I'm
asking
> is that I had never had problems with squeezing WAV file (i.e. making them
> shorter). But I once tried to stretch the WAV file in GoldWave sound
editor
> and artifacts such as clicks and distortion were introduced in sound. Any
way
> to do it without artifacts? Thanks.
>
> --Leonid
>
> PS. The WAV file is 16 bit 48kHz.

Just duplicate every 7200th sample.
Anonymous
October 13, 2004 7:40:48 PM

Archived from groups: rec.audio.pro (More info?)

Leonid Makarovsky wrote:
> I have a 2 hour WAV file that I need to make 1 second longer. I was
doing
> the video capture and audio was ahead of video by 1 second factor.
>
> So I need to stretch it without losing quality. The effect should be
the
> same as if I played an analog cassette tape a bit slower. The reason
I'm asking
> is that I had never had problems with squeezing WAV file (i.e. making
them
> shorter). But I once tried to stretch the WAV file in GoldWave sound
editor
> and artifacts such as clicks and distortion were introduced in sound.
Any way
> to do it without artifacts? Thanks.
>
> --Leonid
>
> PS. The WAV file is 16 bit 48kHz.

Hi Leonid.

You specified that the audio is ahead by 1 second. So, are you sure you
need to stretch the audio? Do you mean that the audio starts off
in-sync but that by the end of the playout the audio is ahead by a
second? Otherwise all you need to do is re-align the audio with the
video.

If you, indeed, you need to stretch the audio by 1 second then that
amounts to .013% change (very little for sure). I have used the the
time-stretch facilities in ProTools, Nuendo and Logic. And each should
be able to perform the change inaudibly (as long as you specify the
highest quality setting vs operational speed).

Regards,
Hassan
Anonymous
October 13, 2004 9:42:07 PM

Archived from groups: rec.audio.pro,rec.video.desktop (More info?)

"Leonid Makarovsky" <venom@cs.bu.edu> wrote in message
news:ckhssg$nhg$1@news3.bu.edu...
> In rec.audio.pro Bob Cain <arcane@arcanemethods.com> wrote:
> : SRC will give an aritfact free result but the the frequency
> : content of the material is also scaled. If you want the
> : pitch to remain unchaged, SRC is not the way to do it.
>
>
> I don't want pitch to remain unchanged. The problem is that SSRC will not
> convert a WAV file with 47999Hz to 48000Hz. It says that it must be
> dividable
> by 2 or 3.

Well... What would it do? Mathematically it would only be adding ONE sample
per second.


>
> So I guess I'm stuck with SoundForge 5.0. I don't like how GoldWave does
> it.
>
> --Leonid
Anonymous
October 14, 2004 10:36:30 AM

Archived from groups: rec.audio.pro (More info?)

"Leonid Makarovsky" wrote ...
> Mike T. wrote:
> :>Cut the audio track apart at several natural pauses and slip it back
> :>into sync so that the error doesn't accumulate.
> :>
> : And fill those stretch marks with "room sound", the background noise
> : you recorded when nobody was speaking.
>
> To be precisely correct it's not even a second I have to stretch this file
> more but .155 of a second. Sounds like a small thing, however, for the
> synchronizing audio and video it's 5 frames of NTSC video and it will
> be noticeable. So really the best thing for me is just to make audio
> slower to match it.

To that casual observer, this almost sounds "obsessive". Half the audio
that goes out over DBS satellite these days has audio that is more out of
sync than that.

I really, REALLY think that fooling around with the timebase, sampling
rate, "stretching" etc is the VERY HARD way to do this. It is SO MUCH
easier to slip pieces of the sound track back into sync at appropriate
points in the NLE that any other method just seems silly, at least IMHO.
Anonymous
October 14, 2004 5:55:56 PM

Archived from groups: rec.audio.pro (More info?)

Hassan Davis <hassandavis@gmail.com> wrote:
: You specified that the audio is ahead by 1 second. So, are you sure you
: need to stretch the audio? Do you mean that the audio starts off
: in-sync but that by the end of the playout the audio is ahead by a
: second? Otherwise all you need to do is re-align the audio with the
: video.

I *assume* that the audio starts in-sync. Honestly, there's no way I can check
that. I can, however, check the audio relative to video position at the end of
the whole thing by re-capturing just the very end of the program, then finding
the same video frame and from that point of time copying audio.

--Leonid
Anonymous
October 14, 2004 5:55:57 PM

Archived from groups: rec.audio.pro (More info?)

In article <ckm0hc$kbe$2@news3.bu.edu>,
Leonid Makarovsky <venom@cs.bu.edu> wrote:
>Hassan Davis <hassandavis@gmail.com> wrote:
>: You specified that the audio is ahead by 1 second. So, are you sure you
>: need to stretch the audio? Do you mean that the audio starts off
>: in-sync but that by the end of the playout the audio is ahead by a
>: second? Otherwise all you need to do is re-align the audio with the
>: video.
>
>I *assume* that the audio starts in-sync. Honestly, there's no way I can check
>that. I can, however, check the audio relative to video position at the end of
>the whole thing by re-capturing just the very end of the program, then finding
>the same video frame and from that point of time copying audio.

If you can't check it, then I assume there is nothing up front that _needs_
to be synched. If that's the case, just shift everything five frames and
let the beginning be out of synch. If there is no dialogue and no synched
effects, it doesn't matter if you're a couple frames off.

If there is dialogue and synched effects up front, then you certainly can tell
if it starts in synch.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Anonymous
October 15, 2004 5:30:18 PM

Archived from groups: rec.audio.pro (More info?)

Scott Dorsey <kludge@panix.com> wrote:
: If you can't check it, then I assume there is nothing up front that _needs_
: to be synched.

I'm not sure about that. All I know for sure that audio vs video in the
beginning differs from audio vs video at the end. But I can't check precisely.
It looks and sounds ok to me. But the beginning is as important as the end.

By the way, I don't worry about the pitch even if I had adjust it more than
even 10 seconds 'cause theoretically if soundcard's internal clock is really
off, I will get correct pitch by stretching or squeezing the file.

--Leonid
Anonymous
October 16, 2004 1:34:10 AM

Archived from groups: rec.audio.pro (More info?)

On 14 Oct 2004 13:55:56 GMT, Leonid Makarovsky <venom@cs.bu.edu>
wrote:

>I *assume* that the audio starts in-sync. Honestly, there's no way I can check
>that. I can, however, check the audio relative to video position at the end of
>the whole thing by re-capturing just the very end of the program, then finding
>the same video frame and from that point of time copying audio.

Where is the first place that you can tell whether audio and video are
in synch? If you drag the audio into alignment here, how far off is
synch at the end (or at the last point where it is an issue)?

CubaseFAQ www.laurencepayne.co.uk/CubaseFAQ.htm
"Possibly the world's least impressive web site": George Perfect
Anonymous
October 18, 2004 6:45:28 AM

Archived from groups: rec.audio.pro (More info?)

Laurence Payne <l@laurenceDELETEpayne.freeserve.co.uk> wrote:
: Where is the first place that you can tell whether audio and video are
: in synch? If you drag the audio into alignment here, how far off is
: synch at the end (or at the last point where it is an issue)?

I'm not sure I understand the question. The audio is out of synch relative
to video.

--Leonid
Anonymous
October 18, 2004 1:52:20 PM

Archived from groups: rec.audio.pro (More info?)

Leonid Makarovsky <venom@cs.bu.edu> wrote:
>Laurence Payne <l@laurenceDELETEpayne.freeserve.co.uk> wrote:
>: Where is the first place that you can tell whether audio and video are
>: in synch? If you drag the audio into alignment here, how far off is
>: synch at the end (or at the last point where it is an issue)?
>
>I'm not sure I understand the question. The audio is out of synch relative
>to video.

At the first shot on the reel, the audio is in synch. At the last shot on
the reel, the audio is out of synch.

If you run the tape, how far in do you first notice that it's out of synch?
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Anonymous
October 18, 2004 5:16:43 PM

Archived from groups: rec.audio.pro (More info?)

In article <ckvao8$kt5$1@news3.bu.edu> venom@cs.bu.edu writes:

> I'm not sure I understand the question. The audio is out of synch relative
> to video.

It may be difficult to descrbe without us being able to see and hear
what you have, but the question is really "what's obvious?" Do you see
lips moving when you hear no speech? Do you see a car crash and hear
the noise a couple of tenths of a second later? Is the wrong
background music playing?

Generally synchronization isn't required for every frame, just the
ones where our brain ties visual and sonic events together. Fix those
by editing so they'll be perfect and don't worry about things that you
hear and can't see (or vice versa).

Otherwise you're just doing an exercise in technology which won't
guarantee the results you need.


--
I'm really Mike Rivers - (mrivers@d-and-d.com)
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me here: double-m-eleven-double-zero at yahoo
Anonymous
October 18, 2004 8:05:06 PM

Archived from groups: rec.audio.pro (More info?)

Scott Dorsey <kludge@panix.com> wrote:
: At the first shot on the reel, the audio is in synch. At the last shot on
: the reel, the audio is out of synch.

: If you run the tape, how far in do you first notice that it's out of synch?

It's proportional. So in 2 hours it's off by .155 of a sec. In 1 hour it's off
by .0775 of a sec.... Etc...

--Leonid
Anonymous
October 18, 2004 8:05:07 PM

Archived from groups: rec.audio.pro (More info?)

Leonid Makarovsky <venom@cs.bu.edu> wrote:
>Scott Dorsey <kludge@panix.com> wrote:
>: At the first shot on the reel, the audio is in synch. At the last shot on
>: the reel, the audio is out of synch.
>
>: If you run the tape, how far in do you first notice that it's out of synch?
>
>It's proportional. So in 2 hours it's off by .155 of a sec. In 1 hour it's off
>by .0775 of a sec.... Etc...

Right. But how far in do you NOTICE it? How far off can it be, before it is
a problem?

If it's two hours long, there should be PLENTY of places to cut a frame of
audio out here and there to match everything up.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Anonymous
October 18, 2004 8:59:30 PM

Archived from groups: rec.audio.pro (More info?)

In article <cl0pji$ipg$2@news3.bu.edu> venom@cs.bu.edu writes:

> : If you run the tape, how far in do you first notice that it's out of synch?
>
> It's proportional. So in 2 hours it's off by .155 of a sec. In 1 hour it's off
> by .0775 of a sec.... Etc...

Do you notice that it's .0775 seconds out of sync? Do you have to look
at a time scale on the computer to know that it's out of sync or is
there some aural-visual cue that tells you that?

Frankly, I think that if you can see that it's 0.0775 seconds out of
sync, you've got a good imagination. What we're trying to get at is
whether you're seeking an exotic solution for a problem that isn't a
problem in practice.


--
I'm really Mike Rivers - (mrivers@d-and-d.com)
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me here: double-m-eleven-double-zero at yahoo
Anonymous
October 18, 2004 11:16:05 PM

Archived from groups: rec.audio.pro (More info?)

Leonid Makarovsky wrote:

> Scott Dorsey <kludge@panix.com> wrote:
> : At the first shot on the reel, the audio is in synch. At the last shot on
> : the reel, the audio is out of synch.
>
> : If you run the tape, how far in do you first notice that it's out of synch?
>
> It's proportional. So in 2 hours it's off by .155 of a sec. In 1 hour it's off
> by .0775 of a sec.... Etc...

I suppose a filter/plugin to do this doesn't exist (maybe I should write
one), but I would be awefully tempted to want to solve this problem by
throwing out individual samples instead of resampling. If you're off by
0.0775 seconds per hour, that's only 0.00215%, which is not even one
sample per second at the 44100 Hz sample rate and just over one sample at
48000 Hz. I'm sort of a purist, but I doubt it would be easy to notice
when one sample is removed here and there. It's equivalent to editing
out about 0.02 milliseconds of audio.

The code for a plugin to do this would be very simple. For a little
added flair, the plugin could look for periods of silence or quiet
parts (or parts without much high-frequency content or with lots of
high-frequency noise) and throw out more samples during those periods
and fewer at other times.

I should explain that if none of this makes sense in the context of this
discussion (or it has already been discussed), it might be because I
just got back from vacation and haven't been following the thread.
(By the way, anyone here ever do installed audio for a cruise ship?
This particular one might have had a deal with Yamaha, because everything
from pianos to drums to consoles was all Yamaha.)

- Logan
Anonymous
October 19, 2004 2:34:13 AM

Archived from groups: rec.audio.pro (More info?)

Scott Dorsey <kludge@panix.com> wrote:
: Right. But how far in do you NOTICE it? How far off can it be, before it is
: a problem?


I start notice things when it's off by 2 NTSC frames which is 0.06 of a second.


Would you recommend jumping to zero crossing and keep inserting silences?

--Leonid
Anonymous
October 19, 2004 2:34:14 AM

Archived from groups: rec.audio.pro (More info?)

Leonid Makarovsky <venom@cs.bu.edu> wrote:
>Scott Dorsey <kludge@panix.com> wrote:
>: Right. But how far in do you NOTICE it? How far off can it be, before it is
>: a problem?
>
>I start notice things when it's off by 2 NTSC frames which is 0.06 of a second.
>
>Would you recommend jumping to zero crossing and keep inserting silences?

No, if the sound is slipping back, I would recommend cutting out video
frames at transitions instead.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Anonymous
October 19, 2004 2:39:10 AM

Archived from groups: rec.audio.pro (More info?)

Logan Shaw <lshaw-usenet@austin.rr.com> wrote:
: I suppose a filter/plugin to do this doesn't exist (maybe I should write


During the video capture there exists an option to resample audio on the fly.
But I'm not sure it would be a good idea.

There's also an option to shift audio proportionally by number of %. I have
to contact the owner of the software package to find out what he's doing there.
If he just samples audio at say 48001Hz, then I'm good.

--Leonid
Anonymous
October 19, 2004 2:41:15 AM

Archived from groups: rec.audio.pro (More info?)

Mike Rivers <mrivers@d-and-d.com> wrote:
: Do you notice that it's .0775 seconds out of sync? Do you have to look
: at a time scale on the computer to know that it's out of sync or is
: there some aural-visual cue that tells you that?


Believe it or not I do notice if I look carefully. I don't notice .03 of a sec.
But .06 is somewhat noticeable.

--Leonid
Anonymous
October 19, 2004 2:47:30 AM

Archived from groups: rec.audio.pro (More info?)

Mike Rivers <mrivers@d-and-d.com> wrote:
: It may be difficult to descrbe without us being able to see and hear
: what you have, but the question is really "what's obvious?" Do you see
: lips moving when you hear no speech? Do you see a car crash and hear
: the noise a couple of tenths of a second later? Is the wrong
: background music playing?


I do different types of capture. In some sound is important. And in some it's
not that important. However, synch is important everywhere.

Some of my captures are live conerts. The drum beats are definitely noticeable.

Some other captures are classic hockey games where I care more about synch of
the sound than the quality. When a puck hits the stick or a board and the
sound of it comes late or early, it doesn't feel right.

Unfortunately I'm a purist too.

--Leonid
Anonymous
October 19, 2004 10:45:37 AM

Archived from groups: rec.audio.pro (More info?)

In article <cl1h62$m3b$5@news3.bu.edu> venom@cs.bu.edu writes:

> Some of my captures are live conerts. The drum beats are definitely noticeable.
>
> Some other captures are classic hockey games where I care more about synch of
> the sound than the quality. When a puck hits the stick or a board and the
> sound of it comes late or early, it doesn't feel right.

This is exactly the type of application where you should manually
synchronize. You really can't depend on electronics to be that stable
for two hours. Get the first drum beat of the shot lined up and it
will stay in sync very accurately until the shot is over (unless of
course you're making a two hour movie of a drum stick hitting the
head).

Nothing wrong with getting it right where it matters, but trying to
make one adjustment based on the starting and ending times and hoping
that will make synchronization correct for the full time period is
wishful thinking. Life isn't that easy.


--
I'm really Mike Rivers - (mrivers@d-and-d.com)
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me here: double-m-eleven-double-zero at yahoo
Anonymous
October 19, 2004 11:22:37 AM

Archived from groups: rec.audio.pro (More info?)

"Mike Rivers" <mrivers@d-and-d.com> wrote in message
news:znr1098130242k@trad...
>
> In article <cl0pji$ipg$2@news3.bu.edu> venom@cs.bu.edu writes:
>
>> : If you run the tape, how far in do you first notice that it's out of
>> synch?
>>
>> It's proportional. So in 2 hours it's off by .155 of a sec. In 1 hour
>> it's off
>> by .0775 of a sec.... Etc...
>
> Do you notice that it's .0775 seconds out of sync? Do you have to look
> at a time scale on the computer to know that it's out of sync or is
> there some aural-visual cue that tells you that?
>
> Frankly, I think that if you can see that it's 0.0775 seconds out of
> sync, you've got a good imagination. What we're trying to get at is
> whether you're seeking an exotic solution for a problem that isn't a
> problem in practice.

Mr. Makarovsky has been flogging this question now in at
least a couple of newsgroups that I read. Frankly, it has taken
on the tone of a troll. He has ignored at least a dozen suggestions
for the most likely methods of fixing the issue (assuming that it
even exists). if he wants to tweak the length of the sound track
by .002% maybe we should just let him learn for himself.
Anonymous
October 19, 2004 8:57:12 PM

Archived from groups: rec.audio.pro (More info?)

Scott Dorsey <kludge@panix.com> wrote:
: No, if the sound is slipping back, I would recommend cutting out video
: frames at transitions instead.

That's a good idea. I haven't thought about it. I'll try it with my next
capture.

--Leonid
Anonymous
October 19, 2004 9:02:01 PM

Archived from groups: rec.audio.pro (More info?)

Richard Crowley <rcrowley7@xprt.net> wrote:
: Mr. Makarovsky has been flogging this question now in at
: least a couple of newsgroups that I read. Frankly, it has taken
: on the tone of a troll. He has ignored at least a dozen suggestions
: for the most likely methods of fixing the issue (assuming that it
: even exists). if he wants to tweak the length of the sound track
: by .002% maybe we should just let him learn for himself.


Mr. Crowley, what went on in your head
Mr. Crowley, did you talk with the dead - Ozzy 1981...

Anyway, I have no time to troll the newsgroups. I posted it to this and
to rec.video.desktop. If you think it's a troll, just ignore the thread.
FYI, I have previously synched audio and video successfully when audio was
longer, but never done it when audio was shorter. Have a nice day.

--Leonid
!