I have got from a neighbor of me a crashed disk with many ten thousands of mp3 files.
After running a data recovery program, I got thousands of records with only a general filename (FILE00000-FILE35000) without album, title, artist.
Searching for programs that can identify this files, I found Picard.
And yes, with this program I could recover many of these files.
But also many of them not.
What can be the reasons?
A reason, I though of, is that the file is not available in the MusicBrainz database.
For instance, Mozart, Quintet in G minor, K516.
I found also other tag programs with a connect to Discogs, but not suitable for my files was my conclusion.
Still, when searching in Discogs database, I find many hits of these Mozart mp3 files.
And WikiPedia shows as info:
Discogs : 151,200,000 tracks
MusicBrainz : ~28,608,635 tracks
So I want to verify now if my conclusion is right, that not all (my) mp3 files will be found in the MusicBrainz database and therefore Picard will not find info about these mp3 files.
Other programs I have used with no results are MP3tag, Tagscan, MusicBee, Kid3.
If someone knows a good alternative, I will be very greatfull to get info about it.
Finally, I think my English is not perfect, I have learned English at a Dutch school in 1959-1965, my apology for this.
There are several things in play here. Like you said, the information might just not be in MusicBrainz. If this is the case, there’s not much to do, other than maybe wait some time and try again and maybe it’s been added in the meantime.
However, the automated track identification is using AcoustID, so in addition to information being in MusicBrainz, it also requires someone to have had the track on hand, done the AcoustID analysis, and submitted this analysis to AcoustID. Again, if there are no matches in AcoustID for the tracks you can’t identify, there’s not much to do at this point other than maybe wait and see if someone’s submitted them at some point in the future.
Note that if the audio has gotten corrupted as a result of the harddisk crash, it may also be that the AcoustID from your files does not match the AcoustID for the track on record, since the audio is maybe significantly different.
One thing you could maybe do is to listen to the non-identified files and see if you can identify the performer, the work (or even just the composer), or anything else that might give a clue about it is, which you can then use to try and figure out what release it’s from… but depending on the amount of non-identified files, this might be way too much work to be worth it.
Following on from Freso’s suggestion:
If you are playing the tracks then using a internet-based music identification service can be surprisingly successful.
I used Shazam and Soundhound apps on an Android mobile phone.
Method:
Play a track in a somewhat acoustically sheltered environment.
Open Shazam app.
Touch the “Shazam” button and then wait beteen 2 and 30 seconds.
In addition what was written before, one way to increase your hit rate for your specific case is lowering the values for “minimal similarity for file lookups” in Picard Options > Advanced > Matching. Maybe to even zero.
The background is that after Picard has got information for matching tracks from audio fingerprinting it still compares the existing metadata of the files against all found matches. This is to find the best suitable match if there is multiple metadata for the same audio.
But if the similarity between existing metadata in the file and the found metadata is below the minimal similarity threshold Picard will not match at all. If you have at least some metadata this is a safeguard against really bad matches.
But in your case where there is not valid existing metadata at all lowering this threshold can give you better results.
Maybe you try “SongKong”. It uses the MusicBrainz Database with AcousticID fingerprinting, but also tries to find matches in Discogs and MusicBrainz with the information that may already be available in your files (e.g. some tags or from filenames). It has a free preview mode which lets you estimate how much it could do for you and your files.
First of all, my apology, the example of Mozart I gave, after a better search, I found tracks in the MusicBrainz database, but with different length of a track.
Again listening to my tracks, I hear for 1-2 seconds a applause at the end of part 1 and 4 of this Mozart Quintett, so problaby a live concert.
And yes, with Google I found a live concert.
And I have already the filename and title of these mp3 files, only not a album, so in fact this example is not correct for my main problem with the mp3 files FILE0000-FILE35000 without album, title and artist.
I have tried to understand how Picard is doing his job.
I found this info:
Scanning (fingerprinting) files
Instead of using release-oriented and metadata-dependent lookups, Picard can try and tag your files as single files (rather than a cluster) based on their audio fingerprint.
If you select a set of files in the left-hand pane and click “Scan”, Picard will find AcoustIDs for your files and query MusicBrainz to find a track that matches them.
But also:
Fingerprinting
If you select a file or cluster in the Left side of the Picard screen and click Scan, Picard will invoke a program to scan the file and produce a fingerprint that can then be used to look up the file on MusicBrainz.
MusicBrainz currently supports only AcoustID (an Open Source acoustic fingerprinting system created by Lukáš Lalinský) but has previously supported TRM and MusicID PUID.
So, I am confused about the use of the fingerprint.
In the first part I read “Picard will find AcoustIDs for your files” but in the second part “Picard will invoke a program to scan the file and produce a fingerprint that can then be used to look up the file on MusicBrainz.”
Must there be already a fingerprint in my mp3 files?
Using the program MP3 diags, by some files I got the message No ID3V2.3.0 tag found, by other files the message ID3V2 tag doesn’t have an APIC frame.
These messages belong to my files FILE000-FILE100, mostly classic music with no results in Picard.
I will be very happy to read a explanation of the use of the fingerprint in Picard.
I also have tried the understand the meaning of a fingerprint (ID3 tag?).
Acoustic Fingerprints are digital data, which will be generated for every track automatically, the first time you put a file into MusicBrainz Picard. No need for you to do anything. It listens to the music and is therefore the same for any file format, be it lossless Flac or mp3 or whatever. When someone enters the CD into the database for the first time, the Fingerprints are stored alongside the tags he enters and everyone else in the future can find the correct tags when he lets Picard scan his music (because his just generated fingerprints for his un-identified music match the ones stored in the database). You should just drag and drop your complete music folder into Picard (into the left column) and let it do its work. On the right side you can see, which albums Picard has identified. If you are satisfied you can save the tags and you are done.
No, the fingerprint is calculated based on the actual audio in the file. It’s basically like Picard is listening to the actual audio and then looking up the database for matching data.
A small correction here: By default Picard will not automatically generate the fingerprints. Use the “Scan” button to let Picard calculate the fingerprint and search for it in the AcoustID + MusicBrainz databases. Picard can be configured to do this automatically by enabling it in Options > General > Automatically scan all new files.
I have got from a neighbor of me a crashed disk with many ten thousands of mp3 files.
After running a data recovery program, I got thousands of records with only a general filename (FILE00000-FILE35000) without album, title, artist.
Searching for programs that can identify this files, I found Picard.
And yes, with this program I could recover many of these files.
But also many of them not.
If your neighbor subscribes to iTunes Match or Apple Music (or is willing to pay for a subscription), this is another resource you can try. iTunes Match uses acoustic fingerprinting, too, but their database is completely different.
To search for matches, you have to use the iTunes software. When it finds a match, you can replace the old MP3 file with a M4A file that has tags, superior sound quality, and a new file name. However, it is much slower and less efficient than Picard. One reason for the slowness is that it automatically uploads non-matching MP3s to the cloud.
If you decide to try this, you should know that Apple inserts the name of the account holder into the files’ metadata. If I recall correctly, they also insert the account holder’s email address. There are other tools you can use to remove this metadata when the process is complete.
In my experience, iTunes match has always kept whatever tags I already had in the file. It provides its own audio, but doesn’t identify the file. (If you know a way to get it to do that, I’d love to know!)
Ah yes, it’s been a few years, but I’m remembering now that I had a post-processing script for this. iTunes Match adds an identifier to the file header; third-party tools can use this to fetch other metadata. There was a Ruby script that looked up the track ID in regional stores, but that script had awkward dependencies, and it hasn’t been updated in years. I searched around and found several ports, translations, and updates, but I haven’t tried them. Here are a few: