Topic: Wave editing: Lossy or Lossless?

[eclark]

Hey guys, I've got a question about Wave Editing. I did a "Destructive edit" on a huge WAV file that accidentally decreased the amplitude -26dB. Since it was "destructive" there was no undo and I couldn't get that data back.

My question is, recording at 24bit /48kHz, once I've saved the file with a lowered amplitude can I later adjust the amplitude (increase by +35dB) and have no loss in fidelity? Or will decreasing the amplitude have a negative effect on the audio file, unrecoverable by a later increase in amplitude? Furthermore, would the quality be even better for me if I had recorded in 24bit /96kHz, because of the higher sample rate?

[dorrcoq]

First question is - why don't you still have the original file on your recorder?

I would think if you amplified the file by 26 dB you should end up with what you had originally.

[eclark]

First question is - why don't you still have the original file on your recorder?

I do, in several locations actually, but why bother wasting time. :-)

[Gordon]

"wasting time"??  seriously?  it's take how long to transfer from an SD card?  have you done more editing or something??  if not I would redo it for sure.  now in theory as stated raising the gain back up 26db's should take care of it.  but again I would just re-transfer it.

[eclark]

Yes. Seriously To copy a 4GB file from my PCM M10 will take about 30 minutes.. Then I need to get all the track markers exactly, name the tracks, some minor cuts and other adjustments that I can't even remember now... it will take a long time, which to me is not worth it.. especially at 11PM, and when I could easily spend the 30 seconds in CPU time to achieve the same effect.

What I was thinking was that if the amplitude adjustment was purely mathematical, then it would seem that it the wave form would be able to be re-adjust "losslessly", achieving better results at a higher sample rate.. My guess would be that this transformation is not lossless, but something tells me also that it could be lossless.

[Brian Skalinder]

Depending on the s/w you use, you may be able to load the original WAV and use the track markers, etc., you've already put in place.

Have you already resampled and/or dithered? (Keep in mind some s/w will dither automatically with each edit, though this is often configurable.)   If already resampled and/or dithered, best not to resample and/or dither again.

[eclark]

Depending on the s/w you use, you may be able to load the original WAV and use the track markers, etc., you've already put in place.

Have you already resampled and/or dithered? (Keep in mind some s/w will dither automatically with each edit, though this is often configurable.)   If already resampled and/or dithered, best not to resample and/or dither again.

I'm using Magix Samplitude Pro X (free trial). I've only been using it for a few weeks, and the workflow has a pretty high learning curve, so I'm not too sure about exporting a cue sheet, and loading the markers back.

I hadn't resampled/dithered. For Samplitude that part is done when you Export to Wav.

Like I said, I am new to the program, and was in "Destructive Edit" mode, and when I changed the Amplitude the edit was destructive (well, AFAIK, maybe I could have reloaded the project file, or something). I was being dumb, but anyway, that's what made me think up this!

Regardless of the file, I'm not really worried about it, I've already cut, re-sampled, labelled and burned the files for a friend, If we need better results I can go back and reload the file (and I'll have to reload the markers because I had already cut the wav file a bit before I marked it).

I'm just wondering if making adjustments to the wave form are lossy or lossless?

I'm also assuming that this is why you are going to have better results with a higher birate/ sample rate.

Next time I'm recording in 24/96kHz =D

[Gutbucket]

Stupid question, and probably too late, but did you try to 'undo' first?

If that didn't work, as Brian mentioned, it may be possible to get the track markers and possibly an edit list of other adjustments from the current file, and reapply that to the original to keep from having to redo that work.  Probably best to grab the original file again to be safe, if it's anything important. [edit- just saw your last post and you mention you're using Samplitude, so you can keep the project with the track markers open, re-import the original audio file and align it, and the track markers should still line up.]

The deal with the -26dB destructive level change- 

Done destructively it is not lossless (a data concept), but may or may not be inaudible (a hearing thing).
Picture in your mind's eye, an artist's conceptual cross-section rendering that illustrates how the dynamic range of the music file is poplulated- noise at the bottom, music data in the middle, and silent unused bits up top (hopefully, unless the recorder clipped). The total range of that usable space is either 16 or 24bits deep total, depending on the setting used when you recorded.  The actual music information takes up a smaller range of bits that has to fit within that total range.

When the gain change was destructively applied, the range of bits used in the file was shifted down within the total range available.  The bottom 5 bits or so were thrown away and 5 bits or so of zeros were added to the top of the range to make up the difference.  If the quietest sounds burried within in the noise floor of the original recording were originally 26dB (not quite 5bits) or more above the lowest significant bit of the 24bit file, then you should be able to simply increase the level of the file again without audible loss, though technically the data representing the noise at the bottom is different so even if the exact opposite gain change is applied, the change is not 'lossless' in a data sense.  But in that case, the only thing you threw away with the level change was noise at the bottom.  However, if there was less than 26bits of noise padding the bottom of the range of the original file, you could loose some low level details since those got thrown away with that data.  If the result is audible or not in that case depends not only on that amount of 'toe-room' at the bottom (as opposed to 'head-room' at the top) present in the original file, but also on the type of material, it's average level, and the overall range of levels.  If the recording has a lot of low level quiet sections and soft parts within or between numbers, you'll be more likely to hear the loss of details in those parts.  If it's all pretty loud, with noise and cheering between songs, and your recording levels weren't super low to begin with, you probably won't notice the change.

Confusing things, most editing software does it's internal calculations at a higher bit depth than that of the file being worked on.  So even though the file holds 16 or 24 bits, the software is probably working with a larger range internally for it's calculations.  Samplitude does this.  But once you save the file again, it dithers and chops off the extra bits again to whatever depth you specify.  That's usually 16 or 24bits, but it's possible to save the entire 36 or 48 bit working file depth if you want to. I'm not sure how Samp works internally, but because of that, if you can't simply 'undo' and/or don't want to revert to the original file, it's a better bet to reapply the gain before closing the project and saving the file to less than the internal working bit depth.

Similar to working with a higher bit depth for it's internal calculations than that of the file, many editing functions and plugins internally oversample the data when manupulating it.  That means that the benefit of recording at higher bit depths and sampling rates is primarily about what you capture originally and how much extra leaway you use to do that (increased dynamic level leaway with more bits, increased frequency range with more samples) and less about compromising things when editing.. assuming you aren't editing destructively until the final output of the project.

Hope that makes sense.