search

usc-imi / aeo-light

AEO-Light 2 is a new generation of optical sound extraction software developed by the University of South Carolina in close cooperation with Tommy Aschenbach. The project is made possible by the generous support of the National Endowment for the Humanities.
Other
41 stars 11 forks source link

truncated wav files #21

Closed stubblmg closed 5 years ago

stubblmg commented 5 years ago

Wav files that we have created in aeo-light v 2.3.2 are showing up as truncated in both MediaInfo as well as BWF MetaEdit and rawcooked.

MediaInfo output in General section: IsTruncated: Yes

BWF MetaEdit Tech data: Errors: invalid Wave: truncated

I'm trying to figure out a workaround, but it would be ideal if the output direct from AEO-Light was not truncated.

The files play back normally and seem OK.

Has anybody else run into this issue? I am unable to identify a setting in the GUI to prevent the files from appearing truncated.

Thanks, Madison

kieranjol commented 5 years ago

We have also noticed this but have not investigated too much,we assumed it was an issue with files created with older versions of aeolight?

retokromer commented 5 years ago

It’s still the case.

wilsbacher commented 5 years ago

Would you be willing to send us an sample wav?

thx Greg

Sent from my iPhone

On May 30, 2019, at 2:28 PM, Madison Stubblefield notifications@github.com wrote:

Wav files that we have created in aeo-light v 2.3.2 are showing up as truncated in both MediaInfo as well as BWF MetaEdit and rawcooked.

MediaInfo output in General section: IsTruncated: Yes

BWF MetaEdit Tech data: Errors: invalid Wave: truncated

I'm trying to figure out a workaround, but it would be ideal if the output direct from AEO-Light was not truncated.

The files play back normally and seem OK.

Has anybody else run into this issue? I am unable to identify a setting in the GUI to prevent the files from appearing truncated.

Thanks, Madison

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

kieranjol commented 5 years ago

P.s I am sure you've noticed that rawcooked refuses to process these files?

On Thu, 30 May 2019, 20:38 Kieran O Leary, kieran.o.leary@gmail.com wrote:

We have definitely seen this issue but have not investigated much. We assumed it was with older files and not with newer ones?

retokromer commented 5 years ago

@kieranjol Also the current RAWcooked snapshot?

kieranjol commented 5 years ago

We have definitely seen this issue but have not investigated much. We assumed it was with older files and not with newer ones?

stubblmg commented 5 years ago

Thanks for the comments. Here is the output from rawcooked:

C:\RAWcooked_CLI_18.10.1_Windows_x64>rawcooked --check 1 X:\AEO-Light\4K_test_film_opt\4K_test_film_opt_346_1500_DPX

Error: untested 4K_test_film_opt_346_1500_DPX\4K_test_film_opt_346_1500.wav Truncated WAV?. Please contact info@mediaarea.net if you want support of such content.

Here is a link to download the wav file (it will be available for a week): https://we.tl/t-HftvsZe0rg

retokromer commented 5 years ago

Here is a link to download the wav file (it will be available for a week): https://we.tl/t-HftvsZe0rg

cc @JeromeMartinez

JeromeMartinez commented 5 years ago

the file is understood as truncated because the WAV chunk size is wrong (0x46 bytes less than reality), so the end of the bext chunk is not inside the WAV chunk (you also lose some bext data...)

0000000 WAVE (12 bytes)
0000000  Header (12 bytes)
0000000   Name:                                 RIFF
0000004   Size:                                 27720738 (0x01A6FC22)
[...]
1A6FC0C   History:                              A=PCM,F=96000,W=24,M=dual-mono
1A6FC2A ,T=A (70 bytes)
1A6FC2A  Header (8 bytes)
1A6FC2A   Name:                                 ,T=A
1A6FC2E   Size:                                 1278037829 (0x4C2D4F45)
1A6FC32  Unknown:                               (62 bytes)

Note at the end the History (from bext) is cut (bext chunk size is correct but limited by the upper box size), and the History remaining bytes are considered as a new chunk (truncated in that case).

BWF MetaEdit could be updated for fixing such issue, RAWcooked could be updated for handling such buggy file (maybe better to fix with BWF MetaEdit instead...), but it would be better to fix aeo-light with right WAV chunk size for new files.

ziggyphenocious commented 5 years ago

I’ve been making lots of wav files from 4k 16 bit tiff files and have not noticed any truncated files or error messages. Are you using dpx is or tiff as a source?

Craig

Craig Nichols The Packard Humanities Institute

On May 31, 2019, at 12:52 AM, Greg Wilsbacher notifications@github.com wrote:

Would you be willing to send us an sample wav?

thx Greg

Sent from my iPhone

On May 30, 2019, at 2:28 PM, Madison Stubblefield notifications@github.com wrote:

Wav files that we have created in aeo-light v 2.3.2 are showing up as truncated in both MediaInfo as well as BWF MetaEdit and rawcooked.

MediaInfo output in General section: IsTruncated: Yes

BWF MetaEdit Tech data: Errors: invalid Wave: truncated

I'm trying to figure out a workaround, but it would be ideal if the output direct from AEO-Light was not truncated.

The files play back normally and seem OK.

Has anybody else run into this issue? I am unable to identify a setting in the GUI to prevent the files from appearing truncated.

Thanks, Madison

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

retokromer commented 5 years ago

but it would be better to fix aeo-light with right WAV chunk size for new files.

+1

Are you using dpx is or tiff as a source?

With 4K 16 bit DPX I can reproduce the error (I don’t have 4K 16 bit TIFF at home).

Yet, why should this depend on the input format?

sjohnson-imi commented 5 years ago

The WAV chunk size misreporting is fixed in v2.3.3 ; I've just put up executables for Windows and MacOS for v2.3.3

Let me know if the problem persists.

https://github.com/usc-imi/aeo-light/releases/tag/2.3.3

retokromer commented 5 years ago

Thank you very much indeed! I am travelling to Vienna for Orphans and therefore can test only next week. Sorry for that!

retokromer commented 5 years ago

@sjohnson-imi Link on the home page should be updated as well.

kieranjol commented 5 years ago

the file is understood as truncated because the WAV chunk size is wrong (0x46 bytes less than reality), so the end of the bext chunk is not inside the WAV chunk (you also lose some bext data...)

0000000 WAVE (12 bytes)
0000000  Header (12 bytes)
0000000   Name:                                 RIFF
0000004   Size:                                 27720738 (0x01A6FC22)
[...]
1A6FC0C   History:                              A=PCM,F=96000,W=24,M=dual-mono
1A6FC2A ,T=A (70 bytes)
1A6FC2A  Header (8 bytes)
1A6FC2A   Name:                                 ,T=A
1A6FC2E   Size:                                 1278037829 (0x4C2D4F45)
1A6FC32  Unknown:                               (62 bytes)

Note at the end the History (from bext) is cut (bext chunk size is correct but limited by the upper box size), and the History remaining bytes are considered as a new chunk (truncated in that case).

BWF MetaEdit could be updated for fixing such issue, RAWcooked could be updated for handling such buggy file (maybe better to fix with BWF MetaEdit instead...), but it would be better to fix aeo-light with right WAV chunk size for new files.

What would BWF metaedit need to do, just update the value to the correct one?

JeromeMartinez commented 5 years ago

What would BWF metaedit need to do, just update the value to the correct one?

Some heuristic tests for having good confidence this is the pattern of an issue (here: testing that the file is more coherent if level 2 chunk size has priority other level 1 chunk size), then update of the value to the correct one.

sjohnson-imi commented 5 years ago

@sjohnson-imi Link on the home page should be updated as well.

Done.

retokromer commented 5 years ago

@sjohnson-imi FYI: It was reported that the macOS installer crashes, but I cannot verify this now.

sjohnson-imi commented 5 years ago

@sjohnson-imi FYI: It was reported that the macOS installer crashes, but I cannot verify this now.

I cannot replicate. There is no installer. The .dmg contains just the .app

retokromer commented 5 years ago

I cannot replicate. There is no installer. The .dmg contains just the .app

I presume while opening the .dmg, but as said I cannot test it now on my end.

JeromeMartinez commented 5 years ago

It was reported that the macOS installer crashes,

I tested on my macOS 10.12 and no crash. I guess it is specific to a platform version, need more details.

sjohnson-imi commented 5 years ago

I presume while opening the .dmg, but as said I cannot test it now on my end.

OK. The github version of the .dmg opens (and the app runs from the image) in my tests. So let me know what you find.

wilsbacher commented 5 years ago

Confirming that 2.3.3 dmg downloaded and opened without crashing on my MacPro (2018) running Mojave.

On Jun 6, 2019, at 8:59 AM, sjohnson-imi notifications@github.com wrote:

I presume while opening the .dmg, but as said I cannot test it now on my end.

OK. The github version of the .dmg opens (and the app runs from the image) in my tests. So let me know what you find.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/usc-imi/aeo-light/issues/21?email_source=notifications&email_token=ABEVDXQCSXUDZCTYH5FEOEDPZEC2DA5CNFSM4HRIPEF2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODXCYBRI#issuecomment-499482821, or mute the thread https://github.com/notifications/unsubscribe-auth/ABEVDXRSQW2ZLZEQDALHQ2LPZEC2DANCNFSM4HRIPEFQ.

retokromer commented 5 years ago

I confirm that it works on my end as well. It seems to me to be a hardware issue on the reporter’s computer (the ominous Kernel stack memory corruption message).

ziggyphenocious commented 5 years ago

Install worked ok on my Mac Pro running Mojave.

Craig

On Jun 6, 2019, at 6:30 AM, Reto Kromer notifications@github.com wrote:

I confirm that it works on my end as well. It seems to me to be a hardware issue on the reporter’s computer (the ominous Kernel stack memory corruption message).

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.