A la une

Gérer une collection de fichiers MIDI, pourquoi et comment ?

Posséder des milliers de fichiers MIDI, pourquoi pas ? Mais à partir d’un certain nombre, le tri manuel ne suffit plus.
Comment s’assurer de leur unicité ? Comment être sur que l’on ne possède pas déjà ceux que l’on ajoute ?
Karaboss propose des fonctions de détection de doubles, puissantes et efficaces, pour vous épauler dans cette lourde tâche.

Posséder un grand nombre de fichiers, à la base, ça n’a pas grand intérêt si on n’en fait rien.
A part peut être si on a une âme de collectionneur. Mais pourquoi diable collectionner des fichiers MIDI quand on peut collectionner des boites de camembert ?

Bon, déjà il faut avoir le temps de les écouter, ou pire, apprendre à les jouer si on est musicien, ou à les chanter si on est chanteur.

Mais sachez que, de toute façon, vous aurez beau en avoir des quantités astronomiques, vous n’aurez jamais celui qu’on vous demande, ou que vous rêvez d’ajouter à votre répertoire.

Et puis souvent, quand on a le bon fichier, il est en double, en triple, en quadruple etc…
Parmi ces doublons, certains ont une qualité sonore excellente, mais parfois carrément désastreuse.

Voilà, pour ce préambule. L’idée n’est pas de savoir quoi faire de ces 30 000 fichiers, légués par votre grand père, mais de faire en sorte qu’ils soient uniques. Sans compter qu’avoir des doubles d’un même titre peut avoir plusieurs significations.

Des orchestrations différentes d’un même morceau

Dans ce cas, avoir plusieurs versions d’un même morceau présente un intérêt évident.
Version acoustique, version électrique, version live etc…
Si la qualité est au rendez-vous et si vous êtes musicien, ou fan de l’artiste, il peut être intéressant de les conserver.

La même orchestration, mais de qualités très diverses

Dans ce cas il s’agira de conserver la meilleure version.
Le seul moyen d’y parvenir est d’écouter chaque morceau et d’éliminer impitoyablement ceux qui vous paraissent mauvais.
C’est difficile de jeter, je sais, c’est trop définitif.
Qui n’a pas en sa possession les versions (1), (2), (3) voire (4) d’un même morceau ?
C’est malheureusement parfois dû au fait qu’aucune version ne sort vraiment du lot, et dans ce cas pourquoi jeter telle ou telle version ?

Et enfin les vrais doubles

Soit ils sont déjà là, tapis insidieusement dans les tréfonds de vos arborescences de répertoires.
Soit ils sont disséminés parmi les fichiers que vous vous préparez à ajouter à votre bibliothèque.

Comment les identifier avec certitude ?
C’est là que les utilitaires proposés par Karaboss deviennent intéressants !

Pour ce faire, Karaboss propose deux modules de recherche :

1. Rechercher les doublons au sein d’une arborescence :
   Cela peut être votre bibliothèque, ou une autre arborescence que vous sélectionnerez.
2. Rechercher les doublons par rapport à votre bibliothèque :
   Il s’agit dans ce cas de comparer les fichiers que vous souhaitez ajouter à ceux de votre bibliothèque pour éliminer ceux que vous avez déjà.

Rassurez-vous, Karaboss ne recherche pas les doubles en se basant sur des critères peu sûrs tels que le nom ou la taille du fichier :
Pour détecter les doublons avec certitude, le logiciel créera un index de tous vos fichiers, en calculant à chaque fois l’empreinte MD5 du fichier (voir plus bas). Si deux empreintes MD5 sont identiques, cela signifie que les deux fichiers ont un contenu identique.

De plus, lorsque des doubles sont détectés, aucune suppression n’est opérée :
A la fin du processus de recherche, vous avez le choix, soit de déplacer les fichiers en double vers un sous répertoire nommé « doubles », soit de ne toucher à rien, mais simplement afficher un listing des doubles vous permettant de traiter chaque fichier manuellement.

Dans le cas du déplacement des doubles vers un sous répertoire, c’est le fichier qui a le nom le plus court qui est déplacé (on ne va quand même pas détruire tout votre travail de renommage !)

Le choix de déplacer les doubles dans un sous répertoire est particulièrement utile lorsque vous voulez importer des nouveaux fichiers dans votre bibliothèque. Vous supprimez ainsi sans effort les fichiers que vous possédez déjà de votre répertoire d’importation, il ne reste plus qu’à traiter les fichiers restants.

Comment procéder ?

Vous accédez aux utilitaires de gestion de fichiers depuis l’explorateur de Karaboss.

Menu « Outils » > « Gérer mes fichiers »

Pour en savoir plus sur MD5 :

L’algorithme MD5, est une fonction qui va calculer une empreinte à partir du contenu d’un fichier. Elle est encore beaucoup utilisée comme outil de vérification lors des téléchargements de fichiers.
Comment ?
C’est une fonction qui calcule le « résumé » d’un texte.
Ce résumé doit être très sensible au texte initial.
Une petite modification du texte doit provoquer une grande modification du résumé.

Le résultat de cette fonction se présente sous la forme d’une simple chaîne de caractères.

L’intérêt d’utiliser les empreintes MD5 pour l’identification des doublons est évident :
Deux fichiers au contenu strictement identiques vont avoir la même empreinte tandis que deux fichiers même légèrement différents vont avoir une empreinte différente.

Exemples d’empreintes MD5 :
Le texte « Au petit bois de Saint-Amand » possède l’empreinte suivante : « 4CF366B73BC4731A527DCC2FBEA0E3D3 »
En modifiant un seul caractère (suppression du tiret de Saint-Amand), cette empreinte change radicalement : « 2AC28BBFCD3011C2E3AAC1334D0D01EF »

STRUCTURE D’ENREGISTREMENT DES FICHIERS CSV

Source :  John Walker (site fourmilab)
Each record in the CSV representation of a MIDI contains at least three fields:

Track

Numeric field identifying the track to which this record belongs. Tracks of MIDI data are numbered starting at 1. Track 0 is reserved for file header, information, and end of file records.

Time

Absolute time, in terms of MIDI clocks, at which this event occurs. Meta-events for which time is not meaningful (for example, song title, copyright information, etc.) have an absolute time of 0.

Type

Name identifying the type of the record. Record types are text consisting of upper and lower case letters and the underscore (“_”), contain no embedded spaces, and are not enclosed in quotes. Karaboss ignores upper/lower case in the Type field; the specifications “Note_on_c”, “Note_On_C”, and “NOTE_ON_C” are considered identical.

Records in the CSV file are sorted first by the track number, then by time. Out of order records will be discarded with an error message from Karaboss. Following the three required fields are parameter fields which depend upon the Type; some Types take no parameters. Each Type and its parameter fields is discussed below.

Any line with an initial nonblank character of “#” or “;” is ignored; either delimiter may be used to introduce comments in a CSV file. Only full-line comments are permitted; you cannot use these delimiters to terminate scanning of a regular data record. Completely blank lines are ignored.

File Structure Records

0, 0, Header, format, nTracks, division

The first record of a CSV MIDI file is always the Header record. Parameters are format: the MIDI file type (0, 1, or 2), nTracks: the number of tracks in the file, and division: the number of clock pulses per quarter note. The Track and Time fields are always zero.

0, 0, End_of_file

The last record in a CSV MIDI file is always an End_of_file record. Its Track and Time fields are always zero.

Track, 0, Start_track

A Start_track record marks the start of a new track, with the Track field giving the track number. All records between the Start_track record and the matching End_track will have the same Track field.

Track, Time, End_track

An End_track marks the end of events for the specified Track. The Time field gives the total duration of the track, which will be identical to the Time in the last event before the End_track.

File Meta-Events

The following events occur within MIDI tracks and specify various kinds of information and actions. They may appear at any time within the track. Those which provide general information for which time is not relevant usually appear at the start of the track with Time zero, but this is not a requirement.

Many of these meta-events include a text string argument. Text strings are output in CSV records enclosed in ASCII double quote (« ) characters. Quote characters embedded within strings are represented by two consecutive quotes. Non-graphic characters in the ISO 8859-1 Latin-1 set are output as a backslash followed by their three digit octal character code. Two consecutive backslashes denote a literal backslash in the string. Strings in MIDI files can be extremely long, theoretically as many as 2²⁸−1 characters; programs which process MIDI CSV files should take care to avoid buffer overflows or truncation resulting from lines containing long string items. All meta-events which take a text argument are identified by a suffix of “_t”.

Track, Time, Title_t, Text

The Text specifies the title of the track or sequence. The first Title meta-event in a type 0 MIDI file, or in the first track of a type 1 file gives the name of the work. Subsequent Title meta-events in other tracks give the names of those tracks.

Track, Time, Copyright_t, Text

The Text specifies copyright information for the sequence. This is usually placed at time 0 of the first track in the sequence.

Track, Time, Instrument_name_t, Text

The Text names the instrument intended to play the contents of this track, This is usually placed at time 0 of the track. Note that this meta-event is simply a description; MIDI synthesisers are not required (and rarely if ever) respond to it. This meta-event is particularly useful in sequences prepared for synthesisers which do not conform to the General MIDI patch set, as it documents the intended instrument for the track when the sequence is used on a synthesiser with a different patch set.

Track, Time, Marker_t, Text

The Text marks a point in the sequence which occurs at the given Time, for example « Third Movement ».

Track, Time, Cue_point_t, Text

The Text identifies synchronisation point which occurs at the specified Time, for example, « Door slams ».

Track, Time, Lyric_t, Text

The Text gives a lyric intended to be sung at the given Time. Lyrics are often broken down into separate syllables to time-align them more precisely with the sequence.

Track, Time, Text_t, Text

This meta-event supplies an arbitrary Text string tagged to the Track and Time. It can be used for textual information which doesn’t fall into one of the more specific categories given above.

Track, 0, Sequence_number, Number

This meta-event specifies a sequence Number between 0 and 65535, used to arrange multiple tracks in a type 2 MIDI file, or to identify the sequence in which a collection of type 0 or 1 MIDI files should be played. The Sequence_number meta-event should occur at Time zero, at the start of the track.

Track, Time, MIDI_port, Number

This meta-event specifies that subsequent events in the Track should be sent to MIDI port (bus) Number, between 0 and 255. This meta-event usually appears at the start of a track with Time zero, but may appear within a track should the need arise to change the port while the track is being played.

Track, Time, Channel_prefix, Number

This meta-event specifies the MIDI channel that subsequent meta-events and System_exclusive events pertain to. The channel Number specifies a MIDI channel from 0 to 15. In fact, the Number may be as large as 255, but the consequences of specifying a channel number greater than 15 are undefined.

Track, Time, Time_signature, Num, Denom, Click, NotesQ

The time signature, metronome click rate, and number of 32nd notes per MIDI quarter note (24 MIDI clock times) are given by the numeric arguments. Num gives the numerator of the time signature as specified on sheet music. Denom specifies the denominator as a negative power of two, for example 2 for a quarter note, 3 for an eighth note, etc. Click gives the number of MIDI clocks per metronome click, and NotesQ the number of 32nd notes in the nominal MIDI quarter note time of 24 clocks (8 for the default MIDI quarter note definition).

Track, Time, Key_signature, Key, Major/Minor

The key signature is specified by the numeric Key value, which is 0 for the key of C, a positive value for each sharp above C, or a negative value for each flat below C, thus in the inclusive range −7 to 7. The Major/Minorfield is a quoted string which will be major for a major key and minor for a minor key.

Track, Time, Tempo, Number

The tempo is specified as the Number of microseconds per quarter note, between 1 and 16777215. A value of 500000 corresponds to 120 quarter notes (“beats”) per minute. To convert beats per minute to a Tempo value, take the quotient from dividing 60,000,000 by the beats per minute.

Track, 0, SMPTE_offset, Hour, Minute, Second, Frame, FracFrame

This meta-event, which must occur with a zero Time at the start of a track, specifies the SMPTE time code at which it should start playing. The FracFrame field gives the fractional frame time (0 to 99).

Track, Time, Sequencer_specific, Length, Data, …

The Sequencer_specific meta-event is used to store vendor-proprietary data in a MIDI file. The Length can be any value between 0 and 2²⁸−1, specifying the number of Data bytes (between 0 and 255) which follow.Sequencer_specific records may be very long; programs which process MIDI CSV files should be careful to protect against buffer overflows and truncation of these records.

Track, Time, Unknown_meta_event, Type, Length, Data, …

If Karaboss encounters a meta-event with a code not defined by the standard MIDI file specification, it outputs an unknown meta-event record in which Type gives the numeric meta-event type code, Length the number of data bytes in the meta-event, which can be any value between 0 and 2²⁸−1, followed by the Data bytes. Since meta-events include their own length, it is possible to parse them even if their type and meaning are unknown.Karaboss will reconstruct unknown meta-events with the same type code and content as in the original MIDI file.

Channel Events

These events are the “meat and potatoes” of MIDI files: the actual notes and modifiers that command the instruments to play the music. Each has a MIDI channel number as its first argument, followed by event-specific parameters. To permit programs which process CSV files to easily distinguish them from meta-events, names of channel events all have a suffix of “_c”.

Track, Time, Note_on_c, Channel, Note, Velocity

Send a command to play the specified Note (Middle C is defined as Note number 60; all other notes are relative in the MIDI specification, but most instruments conform to the well-tempered scale) on the given Channelwith Velocity (0 to 127). A Note_on_c event with Velocity zero is equivalent to a Note_off_c.

Track, Time, Note_off_c, Channel, Note, Velocity

Stop playing the specified Note on the given Channel. The Velocity should be zero, but you never know what you’ll find in a MIDI file.

Track, Time, Pitch_bend_c, Channel, Value

Send a pitch bend command of the specified Value to the given Channel. The pitch bend Value is a 14 bit unsigned integer and hence must be in the inclusive range from 0 to 16383. The value 8192 indicates no pitch bend; 0 the lowest pitch bend, and 16383 the highest. The actual change in pitch these values produce is unspecified.

Track, Time, Control_c, Channel, Control_num, Value

Set the controller Control_num on the given Channel to the specified Value. Control_num and Value must be in the inclusive range 0 to 127. The assignment of Control_num values to effects differs from instrument to instrument. The General MIDI specification defines the meaning of controllers 1 (modulation), 7 (volume), 10 (pan), 11 (expression), and 64 (sustain), but not all instruments and patches respond to these controllers. Instruments which support those capabilities usually assign reverberation to controller 91 and chorus to controller 93.

Track, Time, Program_c, Channel, Program_num

Switch the specified Channel to program (patch) Program_num, which must be between 0 and 127. The program or patch selects which instrument and associated settings that channel will emulate. The General MIDI specification provides a standard set of instruments, but synthesisers are free to implement other sets of instruments and many permit the user to create custom patches and assign them to program numbers.

Apparently, due to instrument manufacturers’ skepticism about musicians’ ability to cope with the number zero, many instruments number patches from 1 to 128 rather than the 0 to 127 used within MIDI files. When interpreting Program_num values, note that they may be one less than the patch numbers given in an instrument’s documentation.

Track, Time, Channel_aftertouch_c, Channel, Value

When a key is held down after being pressed, some synthesisers send the pressure, repeatedly if it varies, until the key is released, but do not distinguish pressure on different keys played simultaneously and held down. This is referred to as “monophonic” or “channel” aftertouch (the latter indicating it applies to the Channel as a whole, not individual note numbers on that channel). The pressure Value (0 to 127) is typically taken to apply to the last note played, but instruments are not guaranteed to behave in this manner.

Track, Time, Poly_aftertouch_c, Channel, Note, Value

Polyphonic synthesisers (those capable of playing multiple notes simultaneously on a single channel), often provide independent aftertouch for each note. This event specifies the aftertouch pressure Value (0 to 127) for the specified Note on the given Channel.

System Exclusive Events

System Exclusive events permit storing vendor-specific information to be transmitted to that vendor’s products.

Track, Time, System_exclusive, Length, Data, …

The Length bytes of Data (0 to 255) are sent at the specified Time to the MIDI channel defined by the most recent Channel_prefix event on the Track, as a System Exclusive message. Note that Length can be any value between 0 and 2²⁸−1. Programs which process MIDI CSV files should be careful to protect against buffer overflows and truncation of these records.

Track, Time, System_exclusive_packet, Length, Data, …

The Length bytes of Data (0 to 255) are sent at the specified Time to the MIDI channel defined by the most recent Channel_prefix event on the Track. The Data bytes are simply blasted out to the MIDI bus without any prefix. This message is used by MIDI devices which break up long system exclusive message into small packets, spaced out in time to avoid overdriving their modest microcontrollers. Note that Length can be any value between 0 and 2²⁸−1. Programs which process MIDI CSV files should be careful to protect against buffer overflows and truncation of these records.

EXAMPLES

The following CSV file defines the five-note motif from the film Close Encounters of the Third Kind using an organ patch from the General MIDI instrument set. When processed by Karaboss and sent to a synthesiser which conforms to General MIDI, the sequence will be played.

0, 0, Header, 1, 2, 480
1, 0, Start_track
1, 0, Title_t, "Close Encounters"
1, 0, Text_t, "Sample for MIDIcsv Distribution"
1, 0, Copyright_t, "This file is in the public domain"
1, 0, Time_signature, 4, 2, 24, 8
1, 0, Tempo, 500000
1, 0, End_track
2, 0, Start_track
2, 0, Instrument_name_t, "Church Organ"
2, 0, Program_c, 1, 19
2, 0, Note_on_c, 1, 79, 81
2, 960, Note_off_c, 1, 79, 0
2, 960, Note_on_c, 1, 81, 81
2, 1920, Note_off_c, 1, 81, 0
2, 1920, Note_on_c, 1, 77, 81
2, 2880, Note_off_c, 1, 77, 0
2, 2880, Note_on_c, 1, 65, 81
2, 3840, Note_off_c, 1, 65, 0
2, 3840, Note_on_c, 1, 72, 81
2, 4800, Note_off_c, 1, 72, 0
2, 4800, End_track
0, 0, End_of_file