Topic: file recovered with PhotoRec...but it's a mess

[junkyardt]

long story here, sorry...so i taped the other night mics and SBD, and for one of the SBD sets, thinking I might need the full internal memory of my LS-10 for the main act, I foolishly switched over to the SD card for the opener. i say foolishly because i've never been able to get this card to work right with my recorder. ever since i bought it A) it doesn't show the right amount of recording time available (it shows the right amount of space available in GB, but the time is way off from what it should be) and B) i've never been able to format it in the recorder; it always gives me an error message. so basically, i've owned it for a while now but have never actually used it to record due to these problems, always opting for the internal memory which has never failed me. so i knew this card had some compatibility issues with the recorder, but still, it did show plenty of recording time available (even though not the right amount). so i figured it can't hurt to finally try to record with it for a 40 minute opening set. well, i set it running, the levels looked good and it was recording the set with no problems. i stopped it at the end, but when i went to play the file, instead of just playing immediately as it would off the internal memory, it gave me a message that it was busy searching/accessing the file or something like that. then when that message cleared, it said there were no files. the file recorded and saved fine, but disappeared immediately upon playback. bizarre.

i tried a bunch of recovery programs with no luck (all would either find nothing or only find a WAV file of size 1kb, which doesn't help me.) then after seeing PhotoRec recommended on here, i tried it and it worked! it scanned every sector on the card for like 45 mins and recovered a 420 MB file that none of the other programs would find (the right size for a 40 minute set at 16/44). PhotoRec said the file was RIFF format, which i was unfamiliar with, though it appears this is simply the container format for the data chunks that make up a WAV file. well the file sounds fine but there's a problem -- there are gaps of silence every so often. actually it's far more precise than every so often. it appears 100% of the data is there, nothing is missing, it's just not seamlessly put together. i'm going through with CEP to zoom in and see where these occur and here are a few examples:

4:41.971
4:42.157

music

5:53.302
5:53.488

music

7:04.634
7:04.820

music

8:15.966
8:16.152

the pairs of numbers are the start and end times for gaps of silence, while the space between the end of one gap and the beginning of the next contains the music. as you can see, the math is pretty precise. there seems to be 1:11.146 minutes (or something very close) worth of music, followed by a gap of .186 seconds, followed by 1:11.146 of music again, and on and on. so i'm left to infer that PhotoRec went through the SD card and gathered up all the raw data and attempted to reassemble it into a new WAV file much like gathering up puzzle pieces and putting the puzzle together. only it seems it wasn't able to do it seamlessly, and left a gap of .186 between each RIFF chunk of 1:11.146.

so my question i guess is if anyone else has encountered this before and how to proceed? the only thing i can figure is to use the numbers i'm compiling like above and track it as precisely as i can into music chunks and silence gaps, delete the gaps and assemble all the chunks back into one file, and *hope* that it sounds seamless. but that's going to be a lot of work, and i'm imagining it might not be so seamless and there may be clicks/pops where the chunks are rejoined. and then it gets even thornier as i want to use this SBD for a matrix with the mic source -- will it sync well? what a mess. though i was happy with the fact that PhotoRec worked when no others would, is it possible there are any other programs that will do what PhotoRec did but do it seamlessly?


UPDATE: nvm, i figured out how to fix this rather easily (though it took a long time).

[Brian Skalinder]

Two thoughts:

• Even though the other apps recovered only a 1 kb file, it could simply report the wrong size due to a file header issue -- similar to the not uncommon 0 kb file problem.  So you might try attempting the same resolution on the 1 kb file as one performs on the 0 kb file.  I think there's a link in the sticky thread at the top of the forum.
• Try cutting out 3-4 sections of silence and see if the WAV recompiles seamlessly.  If it works well, it won't take that long to split and rejoin the 40min or so.  If it's still a mess, no point bothering with the rest, obviously.  Though if it's very close or is basically seamless except for artifacts like pops, you might try very short crossfades and/or enforce your splits at 0-crossings (I forget the proper name...the point where the waveform crosses the horizontal axis; re-joining waveforms when the end of the first portion doesn't align on the horizontal axis with the start of the second portion can cause pops, so forcing splits at 0-crossings ensures the end of the first portion matches the second portion on the horizontal axis).