Improve shaky audio

[Filipon]

Hello,

First of all, excuse my bad English, I am writing to you via a translator (English not being my native language).

There are times when I'm digitizing VHS or listening to old DVDs or obviously the audio hasn't been restored that I encounter some kind of jitter effect on the audio.
The character seems to shake when he speaks, which is quite strange.
I would like to know if there is a solution to fix this problem on Audacity or on another software.

Thanks.

[johns0]

Doesn't sound like you can fix it cause its video/sound issue,best you can do is upload a short clip to this site to see the problem.

[Filipon]

Hello,
thank you for your reply. Attached is an excerpt where you can hear the tremor effect on the vocals.

[Cornucopia]

It sounds like your source (VCR?) had a slight loss of sync and then recovered, and during the loss of sync, the speed of the (tape?) playback varied.

When playback varies like that, what results is usually a frequency modulation, aka vibrato/warble/flutter.

If the vibrato frequency were known, and was a steady, constant, pure single frequency-single amplitude oscillation, it would be fairly easy to counteract it by applying the inverse signal to modulate it.
So, for example, if content signal had 50Hz tone and it varied by exactly 5Hz for 5 seconds, and the resulting signal became 45-55Hz, applying an inverse signal (same oscillation, phase inverted) of 5Hz would raise back up the 45Hz to 50 and lower the 55Hz to 50, and restore the original signal.

HOWEVER, because here the frequency is unknown, and varies in strength over time, and probably varies in frequency over time, attempting to apply an inverse signal would do more harm than good, as if it isn't exactly matching and inverted, you could/would likely double the amount of vibrato (in sections) rather than remove it.

Other than trying to re-capture the clip (with a copy that doesn't have loss of sync, and so remains steady in playback), there is nothing you can do.


Scott

[Filipon]

Hello and thank you for your answer.

Indeed, the source here comes from a VHS PAL but I have already noticed exactly the same problem on commercial DVDs published about fifteen years ago.

If there had been a loss of synchronization, I suppose it should have been visible on the video, right? If necessary, when I watch the video, I do not notice any desynchronization, the voices are perfectly aligned with the movement of the lips.
Also, the voice-shaking effect is consistent throughout the episode.

[Cornucopia]

If the loss of sync is bad enough, yes, it would show up as dropped frames/freezing, etc in video...unless you had a framesync or whole frame-based TBC. What would improve a capture like this is using a VCR with an internal line-based TBC that has control of the tape speed as part of its feedback loop.
However, I guess it IS possible that there is a line TBC but it was acting up and overcompensating?
There is also the possibility that the issue is seen only in the audio, due to the fact that the audio head is further down/up the path of the tape travel. Capstan pinch roller irregularities?
And if the warble/flutter is minor, it may be being compensated for on the video side, but not on the audio side. Flutter isn't necessarily an overall shift in the timing, just micro-shifts back & forth.

BTW, "sync" that I was referring to is NOT the synchronization between the audio & the video, but the separate sync pulse track on the tape. When it loses that sync, it usually "freewheels" for a while until it can get it again, so it may stay quite close to the expected speed (and exhibit no issues on the video side), but even a slight variation is quite noticeable on the audio side. That's just how our ears are designed.

Regardless, other than attempting recapture with a better device chain, there is NOTHING you can do about it.

Commercial DVDs may have also been captured with issues, and just left that way through the encoding, authoring, distribution phases.
VERY MUCH depends on the title, but it is also possible that the flutter is baked into the original mix. If you can find and compare a section that has sustained music behind it, you can tell further: if it flutters the music, it is likely the released tape capture that is at issue, if it flutters only the voice but NOT the music, it is baked into the master mix.


Scott