I have a corrupt .wav file that I am trying to fix. I read many posts and saw tutorials on youtube about using Audacity to fix the problem, however I didn’t find one that applied to my situation exactly.
I recorded 2 .wav files, one is working and the other, for some reason, is corrupt and will not play. With the working .wav, I learned the following info:
Codec: IMA WAV ADPCM Audio (ms)
Channels: Stereo
Sample rate: 24000 Hz
Bits per sample: 16
So I tried to open the corrupt .wav file by importing the raw data. However, I could not find IMA ADPCM in the encoding field, just VOX ADPCM. Is this the same thing? Nevertheless, I selected VOX, and inputted the sample rate as 24000HZ. When I imported it, I got soft static and a few mumbles. At least it is better than not hearing anything or pure static…
Am I doing this correctly? Is the problem that I am not able to select the IMA ADPCM encoding? Not sure what it does, but I experimented by changing the Byte order, and Start offset from 0 to 100 to 500. I still don’t get the pure audible wav form.
I have also tried using GoldWave, and WAV saver but nothing works. Still believe that Audacity is the solution to my problem…
I am using Windows 7. and Audacity 2.0.5
Any help will be greatly appreciated! Thank you very much!
I have a corrupt .wav file that I am trying to fix. I read many posts and saw tutorials on youtube about using Audacity to fix the problem, however I didn’t find one that applied to my situation exactly.
Which was? Posters tend to start the movie in the middle leaving out the beginning where you describe how you got the files, who you were shooting and on what.
Has anything like this happened before? What are the possible differences between the recordings? Did the batteries on the portable recorder go flat before you could save the work? Was the damaged one first or second in the series?
To answer your question. I have been doing a lot of interviews lately and was using the recorder to record the conversations. So far I have recorded 10 interviews and 2 of the .wav files came out corrupted. The rest are fine. I am surprised myself because the batteries were full, and I was using the same recorder with the same settings. The brand of the recorder is called PCHOME and I can take a photo of it if you wanted to see it. I will most definitely use a different recorder from now on.
However, because I can’t hold the interview again, I must somehow figure out a way to retrieve the corrupted .wav file. That is why I am freaking out. It doesn’t even have to be high quality, just as long as I can hear the conversation.
What’s the possibility you ran over the WAV data boundary? Let’s see. The first boundary is 2GB and that works out to be…mono recording at 44100…317MB per hour… That’s about six-ish hours.
No, that’s probably not it. You would have had to pause and not stop several days worth of recording and then the whole batch would have gone under, assuming the recorder wasn’t smart enough to catch it. Olympus recorders start a new file if it gets that bad.
Anything at all magic about those two interviews? Where they much longer than normal?
We don’t do Forensics. So putting a show together from shards and fragments is beyond us unless one of the other elves wants to take a shot at it. Do you have any way to post the whole damaged file, like DropBox or other? How big are they?
Audacity does not produce any YouTube tutorials. Unless we tell you a specific YouTube tutorial is good or correct, you’re at the mercy of the person who did the tutorial. They may not even have the version of Audacity you have (if the tutorials are years old, the menus and controls won’t match with current Audacity).
What tool is giving you that information? IMA ADPCM is 4 bits per sample.
The best tool to use for file analysis is “MediaInfo” from MediaInfo - Download MediaInfo for Microsoft Windows . Get the version without installer, because the installer may have malware or adware. MediaInfo would show the IMA ADPCM file as 4-bit. Open the corrupt file in “MediaInfo” and see if it gives you any information.
No. VOX ADPCM is always without file headers (raw) but IMA ADPCM should have headers. Importing IMA ADPCM as raw data and choosing VOX ADPCM encoding will not import correctly.
I am not sure why there is no IMA ADPCM choice in Import Raw Audio, but it is probably due to the choices offered in the libsndfile library we use for import/export of uncompressed files.
Other than that, as Koz says, post one of the corrupt files somewhere. If necessary, record some fragments you don’t mind being public until it produces a corrupt file.
Dear Koz and Gale,
Thank you for responding to me during the weekend.
To answer Koz’s question, there was nothing special with the two files. They were recorded one after the other and I don’t know why the second came out corrupt. Both are around 22min long and size around 30mb.
I would have to buy an Olympus recorder like you suggested. I did upload the two files onto Dropbox:
To answer Gale’s question, I used VLC player to find the info of the working file. But your suggestion of using mediainfo was much more helpful! The corrupt file generated only the size, while the working file generated this:
Format : Wave
File size : 31.4 MiB
Duration : 22mn 51s
Overall bit rate mode : Constant
Overall bit rate : 192 Kbps
Audio
Format : ADPCM
Codec ID : 11
Codec ID/Hint : Intel
Duration : 22mn 51s
Bit rate mode : Constant
Bit rate : 192 Kbps
Channel(s) : 2 channels
Sampling rate : 24.0 KHz
Bit depth : 4 bits
Stream size : 31.4 MiB (100%)
You were right It was 4-bit. Thanks for answering my question. I thought VOX and IMA were not the same thing.
Is there a way I can have it so that IMA ADPCM is a choice when I try to import raw audio? Maybe that is why the import came out wrong. I searched for solutions such as installing plugins or such, but found nothing and I am using Audacity 2.0.5.
Thank you for providing all the links. I decided to use dropbox and here are the 2 links again.
Once again, thank you Koz and Gale for taking the time to help me. Even though I just started using Audacity, I really like it. It is user friendly and no other software has gotten me closer to recovering the .wav file. If you find out what I am doing wrong please let me know and I will share this info so no one has to be frustrated with the same problem!
Hi Koz!
I have no idea why the link doesn’t work. I posted it in the public folder and copied the public link and pasted it here…
Very sorry for this inconvenience.
I Just renamed the file and here are the links again. I even added a shorter link just in case. If they still don’t work, may I use dropsend to send it to your work e-mail? or If you have dropbox, I can invite you to that public folder. Just tell me the name. Whatever is easier for you…
There are two attached digital dumps. One from the good file and one from the bad. It’s no mystery which is which. One has an orderly header and one doesn’t.
That’s it. I’m not a bit jockey. I have no idea what that means, other than one is a show and one isn’t.
You should like the Olympus equipment. I’m searching for a replacement for my tape recorder. Real tape. My recorder conforms to the definition of success. There’s no paint left and you can’t read the words on the controls any more, but it’s still working and has a job right now. I was considering a WS-803 until they discontinued it…
I exported a 24000 Hz stereo IMA ADPCM file of the same length and added its headers to your corrupt headerless file, and then took another copy of your headerless file and added the headers from the “working” file to that. Both files can be imported normally into Audacity but seem to be very distorted.
Both the corrupted file and the “working” one seem to have the same content (as far as can be deciphered). If so you may not need to rescue the corrupted file.
I don’t think Audacity’s RAW import of IMA ADPCM has ever worked (in Audacity 1.2.x that choice is offered but the OK button is greyed out.) I don’t have any apps on my computer that can raw import IMA ADPCM (Goldwave does not have it, although it has more raw import options than Audacity).
The “clean” one is like that, too. The question is whether the clean one and the failed one are of the same show. If they are, then that’s very serious. It means the second show doesn’t exist.
The shows are “run and gun” captures of interviews. I’ve done that. The only requirement is to be able to make out the words in order to retrieve contact and shared personal information. Nobody is going to put this on NPR (I hope).
On the other hand, if you have the only footage of a national event…
Well, SoX has one up on Audacity then by being able to decode that. The “corrupt” file could be 48000 Hz which would make it closer to the length of the “correct” one and the length of my attempt. My attempt did have all the audio that is in Steve’s version.
The “correct” and “corrupt” versions are not the same as I thought - I did not “listen” to enough of the corrupt one - not that anyone would want to listen to that.
Afternoon!
Thanks for everyone’s help!
I read about the digital dumps online and seeing Koz’s attachments of the codes clearly shows the location where things went wrong. They look so intimidating since I don’t know what they mean!
I’ll check to see if I can use the Hz info that you found Steve! Thanks!
Gale, you’re so kind to decode it for me! It is actually kind of audible! Please please tell me what you did? Or point me in the right direction so I can learn?
This is definitely as close as it has ever been to being recovered. <digital hug!>
BTW, I don’t mind at all you guys listening to the interview. It was interviewing Asian Celebrities and how they got into the entertainment business in Malaysia.
-JW
They look so intimidating since I don’t know what they mean!
I think my only idea was to show you one was neat and orderly and clean and the other was just trash on the screen. Also the orderly one has tiny English words here and there that the damaged one doesn’t
WAVfmt (Wave Format)
data
On the good one, you don’t have to be a programming deity to understand pieces of it.
I just downloaded SoX (which has to run at the command-line because it has no interface with menus and buttons), opened a command prompt where sox.exe was, typed
sox -r 48000 -t ima "JULIE WOON02-corrupt file.WAV" decoded.wav
then hit ENTER.
The “-r” species the sample rate as 48000 Hz (if you choose 24000 Hz or don’t specify the sample rate, so that SoX tries 8000 Hz, the file is too long and plays too slowly).
The “-t” gives SoX the clue that the file is encoded using IMA ADPCM.
It’s not going to get much better than that because it is massively distorted. You could try Effect > Amplify… to -20 dB in Audacity then Effect > Clip Fix… but I doubt it would help much.
I would check if you set your recorder to record or save in mono by mistake. It may have just made mono by adding the two stereo channels together without dividing them by 2 (hence the distortion) then resampled to 48000 Hz to keep the file size the same. Or, it makes mono when you ask for a higher sample rate in order to keep the file size the same. Or it’s just broken.
I found that the recovered recording was a little easier to understand after applying the “Equalization” effect using the “Telephone” preset (certainly a bit less painful to listen to )
Good Morning Everyone and hope you had a great July 4th weekend!
The instructions provided by Gale on decoding the wav files were very clear and I was able to duplicate the result. I am just working on tweaking the results. Hopefully it will sound better than a thousand little needles hitting my ear. lol Thank you and I will provide updates as I work on it further this week.
Have a great week!
It doesn’t even have to be high quality, just as long as I can hear the conversation.
