Closed spoo333 closed 4 years ago
I'm not sure why this would be happening. Can you run it in verbose mode on just the offending files and post the complete log here?
Sure, will do. Will be in touch
On Sat, 11 Jan 2020, 00:16 Ryan C. Thompson, notifications@github.com wrote:
I'm not sure why this would be happening. Can you run it in verbose mode on just the offending files and post the complete log here?
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/DarwinAwardWinner/rganalysis/issues/11?email_source=notifications&email_token=ADRAUCIS2OZH3CEZEKMG7BLQ5EFUXA5CNFSM4KD3CA2KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEIVSZHI#issuecomment-573254813, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADRAUCP7IKIPF2ULLIVSZNLQ5EFUXANCNFSM4KD3CA2A .
Attached is a txt file with verbose output for one of the offending files. I can try other files too, however that would take a while as I don't remember any of the others by heart.
On Sat, Jan 11, 2020 at 2:00 AM David Bauer beelze333@gmail.com wrote:
Sure, will do. Will be in touch
On Sat, 11 Jan 2020, 00:16 Ryan C. Thompson, notifications@github.com wrote:
I'm not sure why this would be happening. Can you run it in verbose mode on just the offending files and post the complete log here?
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/DarwinAwardWinner/rganalysis/issues/11?email_source=notifications&email_token=ADRAUCIS2OZH3CEZEKMG7BLQ5EFUXA5CNFSM4KD3CA2KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEIVSZHI#issuecomment-573254813, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADRAUCP7IKIPF2ULLIVSZNLQ5EFUXANCNFSM4KD3CA2A .
Another check in Puddletag and still no replaygain info of any kind visible in the tags
That's odd indeed. It says it's setting the tags, and there's no warnings or errors that would suggest it failed to do so. Anyway, here's a few things you can try to investigate further, if you're so inclined:
I've done some more tests. Re 1: The files are definitely write enabled, i used chmod 777 to make sure and checked in the file browser as well
I also tried the gain tool included in KDE soundKonverter. No luck with that tool either.
Re 2/3: Been doing a lot of tests removing all tags, converting the flac to wav and back to flac, various mp3 comrpession settings, no luck. According to the access time, there is no access.
However, what is odd is that if I open the original flac file in Audacity and then overwrite the original file, Audacity creates a backup file with "-old1" tagged on to the original file name followed by the .flac extension. Running rganalysis then results in both the newly saved and the original file receiving the appropriate track and album gain tags. Once I remove the "-old1" file, the old problem returns as rganalysis then correctly notices that the new file is by itself, ie no longer part of an album and therefore needs a fresh calculation, presumably to strip out the album tag, but that doesn't happen. If I manually delete the album gain tags from the new file, then rganalysis correctly identifies the new file as a single track and comments "skipping previously analyzed ...."
Following on from that I took a random flac file and placed it in the same directory as the file above. i removed all gain tags from both and assigned the same artist and album tags to the random file as the original offending file. Rganalysis correctly calculates track and replay gain values for both!!
Hope my descriptions make sense. Let me know if I can send you any other info. the md5sum bit I skipped considering my discoveries
On Sun, Jan 12, 2020 at 1:04 AM Ryan C. Thompson notifications@github.com wrote:
That's odd indeed. It says it's setting the tags, and there's no warnings or errors that would suggest it failed to do so. Anyway, here's a few things you can try to investigate further, if you're so inclined:
- Double-check that the file isn't marked read-only.
- Remove all tags from the file (do this on a copy). Then calculate the md5sum of the file before and after running rganalysis. Also, look at the file size, and modification time before and after. See if any of them are changing. If none are, then the file isn't being written to at all.
- Decode the flac to wav and then re-encode it, then try running rganalysis on the new flac file. If this works, then there might be some subtle corruption of the original file's tags.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/DarwinAwardWinner/rganalysis/issues/11?email_source=notifications&email_token=ADRAUCMJZCCQ7I4WOOI5AXLQ5JUAXA5CNFSM4KD3CA2KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEIWO2MY#issuecomment-573369651, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADRAUCLGSH6HYDCLTWCLCB3Q5JUAXANCNFSM4KD3CA2A .
been looking at more audio files that refuse the rganalysis treatment. So far they are all single track files. I'll try combining some of these into a fake album as above
On Thu, Jan 16, 2020 at 2:40 PM David Bauer beelze333@gmail.com wrote:
I've done some more tests. Re 1: The files are definitely write enabled, i used chmod 777 to make sure and checked in the file browser as well
I also tried the gain tool included in KDE soundKonverter. No luck with that tool either.
Re 2/3: Been doing a lot of tests removing all tags, converting the flac to wav and back to flac, various mp3 comrpession settings, no luck. According to the access time, there is no access.
However, what is odd is that if I open the original flac file in Audacity and then overwrite the original file, Audacity creates a backup file with "-old1" tagged on to the original file name followed by the .flac extension. Running rganalysis then results in both the newly saved and the original file receiving the appropriate track and album gain tags. Once I remove the "-old1" file, the old problem returns as rganalysis then correctly notices that the new file is by itself, ie no longer part of an album and therefore needs a fresh calculation, presumably to strip out the album tag, but that doesn't happen. If I manually delete the album gain tags from the new file, then rganalysis correctly identifies the new file as a single track and comments "skipping previously analyzed ...."
Following on from that I took a random flac file and placed it in the same directory as the file above. i removed all gain tags from both and assigned the same artist and album tags to the random file as the original offending file. Rganalysis correctly calculates track and replay gain values for both!!
Hope my descriptions make sense. Let me know if I can send you any other info. the md5sum bit I skipped considering my discoveries
On Sun, Jan 12, 2020 at 1:04 AM Ryan C. Thompson notifications@github.com wrote:
That's odd indeed. It says it's setting the tags, and there's no warnings or errors that would suggest it failed to do so. Anyway, here's a few things you can try to investigate further, if you're so inclined:
- Double-check that the file isn't marked read-only.
- Remove all tags from the file (do this on a copy). Then calculate the md5sum of the file before and after running rganalysis. Also, look at the file size, and modification time before and after. See if any of them are changing. If none are, then the file isn't being written to at all.
- Decode the flac to wav and then re-encode it, then try running rganalysis on the new flac file. If this works, then there might be some subtle corruption of the original file's tags.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/DarwinAwardWinner/rganalysis/issues/11?email_source=notifications&email_token=ADRAUCMJZCCQ7I4WOOI5AXLQ5JUAXA5CNFSM4KD3CA2KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEIWO2MY#issuecomment-573369651, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADRAUCLGSH6HYDCLTWCLCB3Q5JUAXANCNFSM4KD3CA2A .
I tried this with two single mp3 files. As a "fake album" rganalysis puts in the correct tags
On Thu, Jan 16, 2020 at 3:03 PM David Bauer beelze333@gmail.com wrote:
been looking at more audio files that refuse the rganalysis treatment. So far they are all single track files. I'll try combining some of these into a fake album as above
On Thu, Jan 16, 2020 at 2:40 PM David Bauer beelze333@gmail.com wrote:
I've done some more tests. Re 1: The files are definitely write enabled, i used chmod 777 to make sure and checked in the file browser as well
I also tried the gain tool included in KDE soundKonverter. No luck with that tool either.
Re 2/3: Been doing a lot of tests removing all tags, converting the flac to wav and back to flac, various mp3 comrpession settings, no luck. According to the access time, there is no access.
However, what is odd is that if I open the original flac file in Audacity and then overwrite the original file, Audacity creates a backup file with "-old1" tagged on to the original file name followed by the .flac extension. Running rganalysis then results in both the newly saved and the original file receiving the appropriate track and album gain tags. Once I remove the "-old1" file, the old problem returns as rganalysis then correctly notices that the new file is by itself, ie no longer part of an album and therefore needs a fresh calculation, presumably to strip out the album tag, but that doesn't happen. If I manually delete the album gain tags from the new file, then rganalysis correctly identifies the new file as a single track and comments "skipping previously analyzed ...."
Following on from that I took a random flac file and placed it in the same directory as the file above. i removed all gain tags from both and assigned the same artist and album tags to the random file as the original offending file. Rganalysis correctly calculates track and replay gain values for both!!
Hope my descriptions make sense. Let me know if I can send you any other info. the md5sum bit I skipped considering my discoveries
On Sun, Jan 12, 2020 at 1:04 AM Ryan C. Thompson < notifications@github.com> wrote:
That's odd indeed. It says it's setting the tags, and there's no warnings or errors that would suggest it failed to do so. Anyway, here's a few things you can try to investigate further, if you're so inclined:
- Double-check that the file isn't marked read-only.
- Remove all tags from the file (do this on a copy). Then calculate the md5sum of the file before and after running rganalysis. Also, look at the file size, and modification time before and after. See if any of them are changing. If none are, then the file isn't being written to at all.
- Decode the flac to wav and then re-encode it, then try running rganalysis on the new flac file. If this works, then there might be some subtle corruption of the original file's tags.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/DarwinAwardWinner/rganalysis/issues/11?email_source=notifications&email_token=ADRAUCMJZCCQ7I4WOOI5AXLQ5JUAXA5CNFSM4KD3CA2KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEIWO2MY#issuecomment-573369651, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADRAUCLGSH6HYDCLTWCLCB3Q5JUAXANCNFSM4KD3CA2A .
And a few more mp3s added to the same "fake" album and again rganalysis completes the analysis.
On Thu, Jan 16, 2020 at 3:21 PM David Bauer beelze333@gmail.com wrote:
I tried this with two single mp3 files. As a "fake album" rganalysis puts in the correct tags
On Thu, Jan 16, 2020 at 3:03 PM David Bauer beelze333@gmail.com wrote:
been looking at more audio files that refuse the rganalysis treatment. So far they are all single track files. I'll try combining some of these into a fake album as above
On Thu, Jan 16, 2020 at 2:40 PM David Bauer beelze333@gmail.com wrote:
I've done some more tests. Re 1: The files are definitely write enabled, i used chmod 777 to make sure and checked in the file browser as well
I also tried the gain tool included in KDE soundKonverter. No luck with that tool either.
Re 2/3: Been doing a lot of tests removing all tags, converting the flac to wav and back to flac, various mp3 comrpession settings, no luck. According to the access time, there is no access.
However, what is odd is that if I open the original flac file in Audacity and then overwrite the original file, Audacity creates a backup file with "-old1" tagged on to the original file name followed by the .flac extension. Running rganalysis then results in both the newly saved and the original file receiving the appropriate track and album gain tags. Once I remove the "-old1" file, the old problem returns as rganalysis then correctly notices that the new file is by itself, ie no longer part of an album and therefore needs a fresh calculation, presumably to strip out the album tag, but that doesn't happen. If I manually delete the album gain tags from the new file, then rganalysis correctly identifies the new file as a single track and comments "skipping previously analyzed ...."
Following on from that I took a random flac file and placed it in the same directory as the file above. i removed all gain tags from both and assigned the same artist and album tags to the random file as the original offending file. Rganalysis correctly calculates track and replay gain values for both!!
Hope my descriptions make sense. Let me know if I can send you any other info. the md5sum bit I skipped considering my discoveries
On Sun, Jan 12, 2020 at 1:04 AM Ryan C. Thompson < notifications@github.com> wrote:
That's odd indeed. It says it's setting the tags, and there's no warnings or errors that would suggest it failed to do so. Anyway, here's a few things you can try to investigate further, if you're so inclined:
- Double-check that the file isn't marked read-only.
- Remove all tags from the file (do this on a copy). Then calculate the md5sum of the file before and after running rganalysis. Also, look at the file size, and modification time before and after. See if any of them are changing. If none are, then the file isn't being written to at all.
- Decode the flac to wav and then re-encode it, then try running rganalysis on the new flac file. If this works, then there might be some subtle corruption of the original file's tags.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/DarwinAwardWinner/rganalysis/issues/11?email_source=notifications&email_token=ADRAUCMJZCCQ7I4WOOI5AXLQ5JUAXA5CNFSM4KD3CA2KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEIWO2MY#issuecomment-573369651, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADRAUCLGSH6HYDCLTWCLCB3Q5JUAXANCNFSM4KD3CA2A .
So, is your conclusion that rganalysis fails to save tags on tracks that don't belong to an album?
My conclusion so far is that rganalysis (and/or the backends like bsgain and audiotools and whatever else does work in the background unbeknownst to me) does not seem to be able to write tags after analysis into a very small number of files of my very large audio file collection, and when I put those untagged files together under one album and artist name, rganalysis can successfully calculate replaygain for those very few offending files. I certainly have audio files that don't belong to albums that rganalysis has successfully computed. I say "has computed" because I've been using many iterations of rganalysis on different Ubuntu and derivative versions over the years with obviously different backends too so I can't say if those audio files that don't belong to albums were successfully calculated by rganalysis recently or quite some time ago. I'll try to do some tests later on with new, as yet unanalysed single track audio files to see if I can narrow the issue down. Will get back to you later when I have access to my system again
Ok, thanks for all your work isolating the issue.
Thanks for appreciating my work. Thank you for providing the script in the first place and for refining it time and time again.
It just occurred to me that introducing new audio material is a daft idea. I'll make a copy of some of the files that I know to have worked previously, strip them off their gain tags and then run the script.
I'll be checking in again later
On Thu, Jan 16, 2020 at 5:40 PM Ryan C. Thompson notifications@github.com wrote:
Ok, thanks for all your work isolating the issue.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/DarwinAwardWinner/rganalysis/issues/11?email_source=notifications&email_token=ADRAUCINTEVEYVEKUBNHUC3Q6CLZJA5CNFSM4KD3CA2KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEJE5MQI#issuecomment-575264321, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADRAUCL2PFO7OHIM7AV2JP3Q6CLZJANCNFSM4KD3CA2A .
Copied previously working audio files, some full albums, some single tracks, stripped of their gain tags. All albums got processed correctly, unfortunately not a single non-album track did. I'll now try a quick install of a different Ubuntu version, possibly 19.10, to see if the results are similar
On Thu, Jan 16, 2020 at 8:32 PM David Bauer beelze333@gmail.com wrote:
Thanks for appreciating my work. Thank you for providing the script in the first place and for refining it time and time again.
It just occurred to me that introducing new audio material is a daft idea. I'll make a copy of some of the files that I know to have worked previously, strip them off their gain tags and then run the script.
I'll be checking in again later
On Thu, Jan 16, 2020 at 5:40 PM Ryan C. Thompson notifications@github.com wrote:
Ok, thanks for all your work isolating the issue.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/DarwinAwardWinner/rganalysis/issues/11?email_source=notifications&email_token=ADRAUCINTEVEYVEKUBNHUC3Q6CLZJA5CNFSM4KD3CA2KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEJE5MQI#issuecomment-575264321, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADRAUCL2PFO7OHIM7AV2JP3Q6CLZJANCNFSM4KD3CA2A .
I wouldn't bother with a different Ubuntu version. I'm guessing that the current version of my script is silently failing on non-album tracks, and any non-album tracks you have that are properly tagged were actually tagged by a previous version that did work. I think that's enough information for me to start debugging on my end.
I'm halfway through the process so I'll finish it for completeness sake :)
Thanks for looking at potential bugs. Much appreciated
On Thu, 16 Jan 2020, 21:50 Ryan C. Thompson, notifications@github.com wrote:
I wouldn't bother with a different Ubuntu version. I'm guessing that the current version of my script is silently failing on non-album tracks, and any non-album tracks you have that are properly tagged were actually tagged by a previous version that did work. I think that's enough information for me to start debugging on my end.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/DarwinAwardWinner/rganalysis/issues/11?email_source=notifications&email_token=ADRAUCMD32WRANMFJZNSPZTQ6DJEDA5CNFSM4KD3CA2KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEJFVTYY#issuecomment-575363555, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADRAUCO5GYCWSZOAQDQUSEDQ6DJEDANCNFSM4KD3CA2A .
You were right. Same result in Ubuntu 19.10
On Thu, 16 Jan 2020, 21:56 David Bauer, beelze333@gmail.com wrote:
I'm halfway through the process so I'll finish it for completeness sake :)
Thanks for looking at potential bugs. Much appreciated
On Thu, 16 Jan 2020, 21:50 Ryan C. Thompson, notifications@github.com wrote:
I wouldn't bother with a different Ubuntu version. I'm guessing that the current version of my script is silently failing on non-album tracks, and any non-album tracks you have that are properly tagged were actually tagged by a previous version that did work. I think that's enough information for me to start debugging on my end.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/DarwinAwardWinner/rganalysis/issues/11?email_source=notifications&email_token=ADRAUCMD32WRANMFJZNSPZTQ6DJEDA5CNFSM4KD3CA2KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEJFVTYY#issuecomment-575363555, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADRAUCO5GYCWSZOAQDQUSEDQ6DJEDANCNFSM4KD3CA2A .
I found the bug. It should be fixed in the latest version.
Just tested it. Working great. Thank you!!
On Tue, Feb 4, 2020 at 5:29 PM Ryan C. Thompson notifications@github.com wrote:
I found the bug. It should be fixed in the latest version.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/DarwinAwardWinner/rganalysis/issues/11?email_source=notifications&email_token=ADRAUCOVVD6LXIGPZJW7KL3RBGQZLA5CNFSM4KD3CA2KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEKYPX5Y#issuecomment-582024183, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADRAUCLN4ZE76BCUFR66VMTRBGQZLANCNFSM4KD3CA2A .
This is more a question than a bug report: I run Linux Mint 19.3, rganalysis works very well, but I have noticed that with some audio files (both flac and mp3) the following happens, no matter if I use audiotools or bsgain.
I've tried Mixxx for some of those files and it does insert gain tags into the file.
Do you an idea what might cause those problems? Thanks