Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: cut and paste/timecode (Read 1706 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

cut and paste/timecode

Hey, I bought this track on juno download and it has a skipping sound every so often but only in the right channel, I thought I'd be able to edit it out somehow with a low pass filter because it's a high pitch skipping noise. My sister had the bright idea of copying the section from the left channel and pasting over the corrupt info in the right channel... +1.

Here's what it looks like in Audacity:



Very curious to me that the skip noise is only positive and not negative... can you explain that and can you tell me what those numbers indicate anyway? the 1, 0, -1 measuring the waveform that is. Also what is the difference between the dark blue and light blue in the wave? Would really appreciate those few questions answered, but here's the real beef of the problem...

I was able to find out how to select the portion of corrupt audio and holding shift and pushing up I could select the same area of the other channel, copy it, shift back down to the corrupt channel and paste the good channel's section over it. I know I need to make sure the beginning and end of the selections are zeroed so it all flows together without abrupt changes in levels.

I think my mistake was setting the sample format to 16-bit right off the bat instead of leaving it at 32-bit float until the project was ready to be exported at which time I could convert to the original format of 16-bit. The reason I think I made a mistake was after a few edits I notice the waveform was being altered even after my highlighted selection... just by a smidgen but I'm trying to keep the file bit-exact aside from my selected edits.

I've heard the Audacity does not work well for exact timing of audio, would you recommend the best program I'd want to use? Possibly Cubase?

Thanks for any help!

cut and paste/timecode

Reply #1 –
(I'm an occasional Audacity user... I'm not  an expert.)

It might be impossible to exactly time-align the waveforms, since the left & right channels are different, and the zero-crossings can be at slightly different times.  And, if you zoom-in far enough, you can see the individual samples and you can see where the zero crossing is.*

There is an option in the Edit menu to Find Zero Crossings.  The timing of your selection is shown at the bottom of Audacity's window.**

Quote
Very curious to me that the skip noise is only positive and not negative...
We don't know, since we don't know what caused the glitch.

Quote
...and can you tell me what those numbers indicate anyway? the 1, 0, -1 measuring the waveform that is.
It's the sample values normalized between +1 and -1...  Or, you can think of it as 100% positive to 100% negative.  With the wavefroms normalized, you can mix & match 8-bit, 16-bit, 24-bit, and 32-bit files.

100% is also referenced as 0dBFS (Zero decibels Full Scale).  For example, with 16-bits, you can count up to 65,535 (decimal).  Or, 16-bit WAV files use one bit for the +/- sign and go from -32,768 to +32,767.  Those numbers represent -100% and +100%.    This tutorial shows how those samples represent the positive or negative "height" of the waveform at any moment in time. 

Quote
I think my mistake was setting the sample format to 16-bit right off the bat instead of leaving it at 32-bit float until the project was ready to be exported at which time I could convert to the original format of 16-bit.
  If the original is 32-bits, you should leave it at 32-bits for editing.    Like most audio editors, Audacity always works in 32-bits internally.

The main advantage to 32-bits is that it's floating-point and you can go over 100%.  You're not dong anything to boost the volume at the moment, so this isn't an issue.  You can also go very-low in level/volume without loosing resolution.  If you start adjusting the volume, or using other effects such as EQ, you might need to be concerned with the peak levels.    In those cases you can allow Audacity to boost the levels above 0dBFS as long as you re-adjust the levels before saving in integer format.

When you're all done, you can save in the format of your choice, but there's not much point in saving in 32-bit, since there are no 32-bit or floating-point audio-DACs, and 16-bit is more than adequate for humans in real-life listening environments.

Quote
...what is the difference between the dark blue and light blue in the wave?
I believe the light blue is the average, and the dark blue is the peak level.  They are trying to give you a good representation of what the waveform "looks like".  It can't show you the exact waveform, because, unless you zoom way-way-in, there are not enough pixels on your screen to show all of the samples that make-up the waveform.   

Quote
I've heard the Audacity does not work well for exact timing of audio...
I'm not sure what they are trying to say...  For example, with 44.1kHz audio, there are 44,100 samples per second.  If you don't move those samples around (bu cutting & pasting, etc.) they don't get moved in-time and the timing does not get changed. 

Really, there is no "timing" in an uncompressed digital audio file...  Just a bunch of sample values.  If we know the sample rate is 44.1kHz, then we know that the 44,101th sample should be exactly at the 1 second mark.    The only way it get's off, is if the clock in your soundcard is off.  But, that's only during recording & playback and has nothing to do with your audio editor.

P.S.
* If you have a DC offset, there may not be an actual zero-crossing near where you need it.  I've found that the easiest way to fix an offset is to apply a ~10 or ~20Hz high-pass filter to the whole file.  If you zoom-in, remember you're not looking for a sample with a value of zero.  You are looking for a  zero-crossing where one sample is negative and the following sample is positive (or vice-versa).

**   You can change settings if you're not getting enough time-resolution.  You can display all the way down to the actual sample number, or HH:MM:SS + sample.

cut and paste/timecode

Reply #2 –
Dang, thanks for the effort Doug! Your post explains a lot. I'll try start my editing over again while keeping it in 32-bit float and using the zero-crossing function. I'll post the original and my edit when I'm done

Thanks again for the in-depth reply!

cut and paste/timecode

Reply #3 –
I just realized I forgot to leave out one of most important parts I was wondering about... is there anyway to set up cues so that as the track plays in Audacity I can just keep pushing a key or clicking a button as I hear the scratch noise so that I can track them easier and not have to do it in such sectional listening?

...or better yet, if Audacity can't track cues like that, does anyone know of a program that will handle multiple cues of an audio file by a trigger such as a key press?

cut and paste/timecode

Reply #4 –
Alright, editing went pretty well... the zero crossing didn't help perfectly as samples in either channel are slightly different, but it's close enough that it's not noticeable at normal playback speed. I found out I could turn the playback speed all the way down to .1 and preview sections at that speed to really here EXACTLY where the scraping noise in the right channel was, worked like a charm. Here's the finished product: http://www.multiupload.com/VG8X49B52B

There may still be some scratching and imperfections such as the difference in zero crossing on either channel but it's damn close to perfect! Thanks for the help Doug!