Is is possible to "analyse only" when licensed and does that preclude writing changes without rerunning scan?

[paultaylor]

Further Analysis of Basic Metadata Inconsistencies

Multiple values for Album Field
Status Report:15 folders
Fix Songs Report:85 folders
So as explained above the increase is due to the fact that Picard had only matched some tracks, so these were then updated by SongKong and the others were not which could lead to differences in Album field, mostly resolved if you rerun with FixSongs set to Rematch

Missing AlbumArtist Field
Status Report:13 folders
Fix Songs Report:13 folders
Actually of the 13 folders that started with missing album artists 11 of them are fixed by Fix Songs but still listed because of the bug I explain above, fixed for next release.

Multiple values for Album Artist Field
Status Report:8 folders
Fix Songs Report:18 folders
So most of the original eight seem to have duplicate names in some songs e.g Keith Jarrett; Keith Jarrett; Keith Jarrett Trio instead of Keith Jarrett; Keith Jarrett Trio and these mostly fixed by Fix Songs.

The eighteen are mostly new and half are because the folder contains duplicates songs, so what happens is SongKong tries to match all the songs into one album and fails because of the duplicates, it then realizes that there appear to be some duplicates so splits into two groups, matches the main group to an album and updates them. It then trys to match the second group which just contain the duplicates but because you have the Only allow match if all tracks in album were matched option enabled and you dont have enough duplicates to completely match the album again it cannot match again so these songs are not updated so then you have differences between these and the matched ones in the album artist field.

If Only allow match if all tracks in album were matched was disabled they would not occur, but useful that they are listed as proper solution is to delete these duplicate files.

The other half are the same issue as the multiple values for Album fields. because Picard had only matched some songs these were the ones updated leading to Album Artist now being different, e.g
Guitar Hits - Guitar Hits Play Eric Clapton

Missing Artist Field
Status Report: 423 folders
Fix Songs Report: 423 folders
In the Status Report these are folders with missing artists, most do not appear to be matched to MusicBrainz. If none of the songs in the folder have a value, there is a small bug in report, it says folders contain 0 songs but none have a value for the Artist field where the 0 should be the number of actual songs, raised a new issue to fix for next release.

Fix Songs fixes some of these, but because of the same bug as found for Missing AlbumArtist field it still lists folders that no longer have issue

Missing Title Field
Status Report: 42 folders
Fix Songs Report: 42 folders

Same issues as Missing Artist Field

Missing Track No Field
Status Report :42 folders
Fix Songs Report: 42 folders

Same issues as Missing Artist Field

[paultaylor]

MusicBrainz Inconsistencies

Incomplete Folders
Status Report:103 folders
Fix Songs Report:129 folders

In Status Report these are a mixture of folders that were matched to an invalid MusicBrainz release that did not match all tracks, or folders containing some duplicate tracks. Many of these folders were listed under Basic Inconsistencies:Multiple Values for Album/Album Artist field section.

In Fix Songs Report some are the same as in Status Report and some are folders containing duplicates whereby we have now matched one set of the files to an album but not the duplicates as explained in the Basic Inconsistencies:Multiple values for Album Artist Field section

Split Folders
Status Report:11 folders
Fix Songs Report:36 folders

The folder contain songs whereby some have been mixed to one version of an album, and some another. The basic metadata is generally consistent (such as album and album artist) so they didnt come up in basic Inconsistencies but actually they are wrongly matched to multiple albums by Picard.

In some cases there are duplicate files

There is also this weird case /qnap/qnap2/P/Phantogram - Phantogram seems to include two different releases but both with the name Phantogram!

Has added an additional 25 folders, these are cases whereby Picard had matched some but not all songs, and then SongKong was able to find an album that matched all the remaining songs. For example /qnap/qnap1/A/Al Stewart - Time Passages contains 31 songs, Picard had already matched tracks 10-31 but not tracks 1-9, but SongKong found an album with nine tracks that matched tracks 1-9. Now Picard should have matched al 31 tracks to the albums it matched 21 tracks to, running with Rematch would fix this.

IncompleteAlbums
Status Report:319 albums
Fix Songs Report:179 albums

In this section we group songs by the value of the album artist and album field rather than folder. The results are a bit misleading because in many cases what has happened is you simply have two versions of the same album and one is matched to MusicBrainz and one is not, for example you have two versions of …But Seriously so it is not really an incomplete album. However, if you using your music player with files grouped by AlbumArtist, Album as is the norm you would see duplicates of each track which is not ideal so probably the Album field of one of the versions should be modified so listed as seperate albums. So it is useful for this issue to be highlighted but probably should not be listed as an incomplete album.

Because Fix Songs was able to match the second version in many cases the count reduces from 319 to 179.

Split Albums
Status Report:37 albums
Fix Songs Report:1614 albums

In Status Report and Fix Songs Report these are again cases where you have two versions of same album in different folder, and both matched to a different MusicBrainz release. So in most cases nothing is actually wrong its just there would be that problem with media player. I think what is needed is only list as split if we have incomplete matches, i.e we have 5 songs in one folder, and 5 songs in another folder both with the same value for album but matched to two different MusicBrainz Release Ids that are only partially matched (i.e have more than 5 songs), need to think about this a bit more.

Because Fix Songs was able to match so may more releases to MusicBrainz the count has increased because different versions being matched to different MusicBrainz releases, hence appear split when not really.

Discs Missing
Status Report:131 MusicBrainz albums
Fix Songs Report:399 Musicbrainz albums

We are now looking from the perspective of the MusicBrainz releases matched to, there seems to be another bug because many listed such as the first one Al Stewart / Time Passages say

MusicBrainz release has 1 discs, but your songs have only been matched to 0 discs

which makes no sense it should only be showing when have matched some but not all discs, raised issue

An example of a valid ones listed is Arch Enemy / As the Stages Burn!
MusicBrainz release has 2 discs, but your songs have only been matched to 1 disc

Its quite common when release has multiple discs that only some are matched because of various reasons

1. You only have some of the discs
2. The extra discs are dvds/non-audio discs that you dont want to rip
3. Dont have the multidisc release but only the multidisc release was availble to match
4. Different discs have been matched to different versions of the release.

After running Fix Songs the number of discs missing has increased, because the number of releases matched to MusicBrainz has increased (from 17% to 85%). It is important to note that although usually SongKong does not allow partial release matches it does allow partial release matches as long as this is restricted to complete matches of discs. This is because of two main reasons:

1. As explained above somtimes users dont want to rip all discs because not all audio (e.g https://musicbrainz.org/release/5ea07cc8-8882-442e-8a73-8e92b30e86a9) , or dont have all discs.
2. Trying to match a complete boxset is difficult because an issue with one file could prevent match, so we allow matching one disc at a time

One Song Matched:
Status Report: 1 MusicBrainz album
Fix Songs Report:36 MusicBrainz albums

Status Report lists only one folder, Fix Songs Report lists some more including

Blackfoot / Marauder

This is also listed under Basic Inconsistencies:Incomplete Folders under both Status Report and Fix Songs Report. It is another example of Picard only matchig one song to a release, and then because of that one song is updated by Fix Songs, however it is not clear to me why it is not listed for the Status Report:MusicBrainz:One Song Matched as well - raised https://jthink.atlassian.net/browse/SONGKONG-2842

Another example is /qnap/qnap2/M/Meghan Patrick - Heart on My Glass in this case there is a duplicate song, so that song is not fully matched to release, but that was already situation before ran Fix Songs so not clear why not listed in the original Status Report.

Songs Missing:
Status Report:71 MusicBrainz albums
Fix Songs Report214 MusicBrainz albums

These are releases where either you are missing some songs or you have duplicates for some tracks.

[paultaylor]

To summarize a few issues with the report, many issues already resolved for next release and hopefuly should have another release out by the end of the week.

[Optimal-Procedure885]

I’m on holiday starting end of week then I should be able to re-engage with all of this. Thanks for the effort.

I’ll eliminate the duplicates (using the md5 of the audio stream to identify instances of the same md5 in the same folder), test both and if both pass delete one. I’ve been meaning to do this for some time but never gotten round to it. This issue is likely a remnant from recovering a drive crash many years back and having to restore / copy from backups…with files having been renamed to reflect tag casing so quite possible for additional copies to have been incorporated due to case differences in file name.i do have a script that detects duplicate and missing track numbers based on metadata. I should add another that looks at file name.

Clearly I have some investigating/cleanup to do. Pleased my library seems to have helped with surfacing some SK anomalies also. Until you pointed out that fix songs shouldn’t be reporting issues it was able to resolve I’d assumed it was intentional.

[paultaylor]

@Optimal-Procedure885

I’ll eliminate the duplicates (using the md5 of the audio stream to identify instances of the same md5 in the same folder), test both and if both pass delete one. I’ve been meaning to do this for some time but never gotten round to it. This issue is likely a remnant from recovering a drive crash many years back and having to restore / copy from backups…with files having been renamed to reflect tag casing so quite possible for additional copies to have been incorporated due to case differences in file name.i do have a script that detects duplicate and missing track numbers based on metadata. I should add another that looks at file name.

We do have a pretty powerful Delete Duplicates task that could do much the same

Enable Find duplicates within the same folder only option
Set Song is a duplicate if has same to Same song and same album(metadata only) and sounds the same option

This will restrict duplicate finding to files within same folder, looks for same artist, title and album metadata so can be used before matched to MusicBrainz with Fix Songs, the sounds the same part uses acoustids which is similar to md5 but can work over different formats and allows minor differences

@Optimal-Procedure885

Clearly I have some investigating/cleanup to do. Pleased my library seems to have helped with surfacing some SK anomalies also. Until you pointed out that fix songs shouldn’t be reporting issues it was able to resolve I’d assumed it was intentional.

It has been very useful, working with different large libraries is very helpful to me.

[paultaylor]

Okay so have just released SongKong 12.0 Technique that fixes the various issues with report, the issue with Add to MusicBrainz and also added a new option to For songs already fully matched called Rematch if Only Partial Match that makes it possible to rematch folders of songs that were half matched by Picard without having to rematch songs that are fully matched.

What would be great is if you could do the following:

• Delete $HOME/.songkong/reports (since going to rerun anyway and prevents later Create Support Files re-uploading these, and quicker than using Delete Reports task)
• Install New Version
• Run Status Report
• Run Fix Songs, ensuring that For songs already fully matched is set to Rematch if Only Partial Match and Only allow match if all tracks in album were matched is disabled for best results
• Upload Create Support Files

Then I can double-check problems have been resolved and you have accurate reports.

Status Report took longer than expected to run on your system, have raised an issue for this but no obvious cause may require substantial reworking to improve. But probably worth explaining that SongKong has no requirement/concept of adding files to a library and so even though all metadata may already be in database we still have to process via folder structure because that is the only way we know if any file has changed or been added since the files were last read into database. But the advantage is we only have to read file last modified date, we don’t have to actually read the contents of the file if the file has not been modified since last read.

I can see the advantage of being able to run Fix Songs in Preview Only and then be able to apply those exact results guaranteed, and quite a few customers have asked if possible - so I am going to prioritize the Apply Last Preview Task issue. But it is not a trivial thing to add and will take some consideration.

[Optimal-Procedure885]

@paultaylor

would be great is if you could do the following

Will do. You’ve been busy!

[Optimal-Procedure885]

Going to kick this off today, there any reason to hold off pending any other updates?

[paultaylor]

No, there won’t be another release before the end of the year.

[Optimal-Procedure885]

@paultaylor

Status Report

Status Report uploaded.

Fix Songs has just been triggered.

[paultaylor]

Great, thanks - nothing major but found a few errors:

1. On Basic Metadata Inconsistencies tab the message

Sort Artist field has no metadata, all songs should have a value for this field not only when the value is different to the 0 field

is incorrect it should say

Sort Artist field has no metadata, all songs should have a value for this field not only when the value is different to the Artist field

Same problem with the Sort Album Artist line, raised issue

2. Basic Inconsistencies:Missing Track No field is incorrectly showing the results for Missing Title instead, raised issue.

3. On the MusicBrainz Inconsistencies:Only Song matched tab it lists

Van Halen / The Very Best of Van Halen and Various Artists / HDtracks 2013 Sampler but when I click on the link

I cannot see any songs matched, then when I click on the folder containing the songs they are all matched to the same release

However, the last three songs don’t have a trackno, and if go to Matched to MusicBrainz/VanHalen we see the album has been split into three

So I think the issue is related to the missing trackno, apart from these two albums all the others listed on the Only Song matched tab are correct, raised issue

4. There is slight issue with the new Multiple Albums tab it seems that some albums that now should be listed here are still in either the Incomplete Albums or Split Albums tab

e.g Corinne Baily Rae Black Rainbows still listed in Incomplete Albums, even though copy fully matched and the other fully unmatched, raised issue

e.g Lorde Virgin still listed in Split Albums even though its two copies both fully matched to a different version of release, raised issue

5. Good to see Status Report only took 2 hours this time, but I don’t know why it was quicker because i have not made any fixes for it, will investigate the old and new logs to see if I can account for difference.

[paultaylor]

Update, all four issues now fixed for next release, but this will not be until next year and should include Apply Last Preview task as well.

Issue 4 can occur if the release being checked is not in the cache so it could happen for any random album, but as demonstrated was only a problem in two cases.

Issue 3 was due to some songs not having tracknos, I have now added extra logic to check for matching tracks by title if no trackno. Additionally Issue 2 meant that the Missing Track No section was missing and so we missed out on a clue since tracks for both the albums that had an issue would have been listed here.

The issues are minor, and only likely to affect very few files.

[Optimal-Procedure885]

If I’m correct in assuming that following ingestion your database contains all metadata it found in the underlying files I think it would be a great boon to SongKong to be able to do an update scan whereby it leverages what’s in its database and only ingests new metadata from new and/or changed files. It’d do away with the need to run such long ingestion scans. It’s no 96% of the way through ingestion, so I should have a Fix Songs upload for you in the next few hours.

[paultaylor]

Whenever you run any task if your files are already in the database and the files have not be modified by another tool since then we always read the metadata from the database rather than the files, but we stil have to traverse the directory structure to identify what files currently exist . And if they have been modified outside of SongKong then the metadata in database would be out of date and would have to read the metadata from the filesystem for those files.

I’m not sure if you are referring to Status Report or Fix Songs or both when you say ingestion but if run Fix Songs after Status Report then it can read metadata from the database as described above, but the main chunk of time taken will be matching the songs to MusicBrainz and Discogs, since only 17% of your songs are matched to MusicBrainz and none are matched to Discogs

If Fix Songs was being run for real not preview and you then ran Fix Songs again the second time round for the already matched songs (now about 85%) because the majority of songs were already matched SongKong would not have to search for matches this time, we would only update from the MusicBrainz/Discogs releases, this is much quicker, epecially since most releases would already be in your cache. It would only attempt rematch of the remaining 15%

If For songs already fully matched is set to Ignore then SongKong woudn’t do anythign with the 85% just try to match the 15%

If Ignore songs previously checked that could not be matched is enabled then SongKong would not try to rematch the 15% that it aready tried to match and failed. It would only try to match any newly added files that it had not seen before.

[Optimal-Procedure885]

Think the support files are with you now.

I’ve just taken a quick gander and feel like the walking wounded.

[Optimal-Procedure885]

I’ve just started browsing through the fix songs report and then navigating to the respective folder to inspect the tags using puddletag. Looking at Basic Metadata Inconsistencies it reports:
There are 97 folders containing songs with different values for the Album field, folders usually represent an album and if so this field should be the same for all songs on the album

The first entry is “/qnap/qnap1/A/Alessandro Simonetto - Erik Satie Works for Piano [2444.1 kHz]”.

My first observation (and not sure whether this carries through to your database and would impact being able make changes to the file tags): The report lists:
/qnap/qnap1/A/Alessandro Simonetto - Erik Satie Works for Piano [2444.1 kHz]

whereas the actual folder name is:
/qnap/qnap1/A/Alessandro Simonetto - Erik Satie Works for Piano [2444.1 kHz]

(two spaces after Piano)

Loading the album into puddletag, selecting all files and entering Extended Tags view see this:

I’ve checked a few others randomly and seeing the same things:

• Where there are two or more consecutive spaces in a folder name you’re eliminating all but one in the report. Not sure if this extends to filenames also
• the sampled albums do not appear to have different values in the ALBUM tag.

Running

SELECT __dirpath
FROM alib
GROUP BY __dirpath
HAVING COUNT(DISTINCT album) > 1 ORDER BY __dirpath;

against my SQLite table yields only 15 entries, of which only 7 are in your list.

[Optimal-Procedure885]

In this tab:

There are a number of instances where you’re detecting a value in an albumartist tag but inspection of the files shows this is not the case:

Querying my db returns the same:

select * from alib where __dirpath='/qnap/qnap2/VA/Singles Gathered' and albumartist is not null;
-- Result: 0 rows returned in 554ms

Also, if an album is explicitly tagged compilation = 1 then its safe to assume there should not be an albumartist entry.

SELECT DISTINCT __dirname
FROM alib
WHERE __dirpath IN ("/qnap/qnap2/P/Peter Maunu - Windham Hill Records Guitar Sampler"
, “qnap/qnap2/VA/Audiophool/JMLab Test CD No. 2 Tools CD”
, “qnap/qnap2/VA/Masters of Metal”
, “qnap/qnap2/VA/Selection of Soul”
, “qnap/qnap2/VA/Singles Gathered”
, “qnap/qnap2/dt/0-AUDIOPHOOL/Prime Test CD”
, “qnap/qnap2/dt/0-VA/MDMS System Conditioning Disc”
, “qnap/qnap2/dt/0-VA/MDMS System Conditioning Disc (2)”
, “qnap/qnap3/tunesdl-backup/0-not-on-alib/2025-09/VA - Cover Songs Vols. I-IV [Mixed Res]”) AND
compilation != 0 order by __dirname;
– Result: 1 rows returned in 605ms
which is “Peter Maunu - Windham Hill Records Guitar Sampler”

[Optimal-Procedure885]

You’ve also forced me into finally running:

SELECT __dirpath, __md5sig, COUNT() AS occurrences
FROM alib WHERE __md5sig IS NOT NULL AND __md5sig != 0
GROUP BY __dirpath, __md5sig
HAVING COUNT() > 1
ORDER BY __dirpath, occurrences DESC;

Sadly it returns many more rows than I’d have liked, some because some albums have consecutive silent tracks (e.g. https://www.allmusic.com/album/when-im-with-you-mw0000632944#trackListing), but most because of human error.

Have deleted over 1000 tracks some being the same album copied into the same folder 2x with different filename conventions, others being case variations on individual tracks, but a lot of that basic stuff is now sorted.

[paultaylor]

@Optimal-Procedure885

The report lists:
/qnap/qnap1/A/Alessandro Simonetto - Erik Satie Works for Piano [2444.1 kHz]

whereas the actual folder name is:
/qnap/qnap1/A/Alessandro Simonetto - Erik Satie Works for Piano [2444.1 kHz]

(two spaces after Piano)

This a quirk of html, if you were to use View Source in your web-browser you would see that the extra space is there

<span>/qnap/qnap1/A/Alessandro Simonetto - Erik Satie Works for Piano  [2444.1 kHz]</span>

but web-browsers collapse multiple spaces into a single space. This is why in your post the example you have posted looks the same because it has removed your extra space.

The only way round it would be to convert the spaces to   but this has its own issues as it would require extra processing and increase size of reports, but its a possibility. This quirk has no effect on actual processing of issues.

Would not help with filename but if you want to find extra spaces in metadata then you can use AutoEdit:Trim

@Optimal-Procedure885

There are 97 folders containing songs with different values for the Album field, folders usually represent an album and if so this field should be the same for all songs on the album

The first entry is “/qnap/qnap1/A/Alessandro Simonetto - Erik Satie Works for Piano [2444.1 kHz]”.
…the sampled albums do not appear to have different values in the ALBUM tag.

So the report is created at the end after main processing of task has completed, so in the Status Report nothing else is done apart from create a report but when you run Fix Songs the report is being applied after all the changes made by the Fix Songs task, but you are only running Fix Songs in Preview mode so nothing has actually been changed. Therefore if you are trying to compare these inconsistencies with your own database method you should be looking at the Status Report, this has 15 differences

Now your followup question maybe why has the number of folders with different values for the Album field increased from 15 to 97 after running Fix Songs, of that im not sure as SongKong is configured to match grouping to one album, I will look further into that.

@Optimal-Procedure885

There are a number of instances where you’re detecting a value in an albumartist tag but inspection of the files shows this is not the case:

Again, issue here is you are looking at the report created by Fix Songs rather than Status Report.

Also there was one unewxpected error

Dec 19, 2025, 5:18:08 PM

could not prepare statement

this is why on summary page it says

before task cancelled by SongKong because of unreported error

[Optimal-Procedure885]

@paultaylor

Again, issue here is you are looking at the report created by Fix Songs rather than Status Report .

So if I’ve understood correctly the Fix Songs Report doesn’t report what it’d do, it’s reporting the aftermath that would be following it having done its thing if it were licensed i.e. any errors and anomalies would be those that remain after the fact?