METHOD FOR FILE IDENTIFICATION AND DEVICE FOR THE IMPLEMENTATION OF THE METHOD.
The invention concerns a method for identifying files, in particular files containing video sequences.
It also concerns a device for implementing this method as well as a recording medium or support containing the identifications.
Here video sequence is understood to mean the different scenes of a professional or amateur film, of a game or even of television broadcasts.
It is commonplace to identify each video sequence by a written instruction (inaudible) on the screen. For example, DVD disks contain a table stating the different parts of the disk so that the user may directly access any part of the disk that interests him/her, if he/she does not want to watch the whole disk. By way of another example, television operators broadcast programmes or guides stating the broadcasts under way and future broadcasts on screen.
Plain titles are not always sufficient to represent the video sequence properly. Furthermore, the austere appearance of a written text may be felt to be forbidding for a video sequence, which is more often than not displayed for leisure or for playing. Furthermore, these titles force the viewer to read and cut down the potential emotion that a scene may make a person experience.
Moreover, the language used is not necessarily understood by all the users.
In order to make the presentation of the video more attractive, it is common practice to accompany the presentation of each sequence by an audio accompaniment, especially a musical one, taken from the pertinent sequence. However, these accompaniments constitute illustrations and not identifications of the video sequences.
The goal of the invention is to provide a method allowing an audio identification, without text, of the content of the files representing processes lasting for a given duration, in particular for video sequences.
For this purpose, the invention envisages associating an audio identification to each file representing the contents of this file, with this identification having a short duration compared to that for the file and being different from the latter. The audio identification is preferably a musical identification. In the event of a video sequence, it is different from the soundtrack for the sequence. Preferentially, the identifications are to be made with the help of an automatic music production device. In this case, the contents of each sequence may be represented by a digital signal controlling the automatic music production device. The duration of the musical identification is, in one instance, at most 20% and preferably less than 10% of the duration of the sequence to be identified.
The audio identification, whether produced or not by a musical composition generating device, is made up by images with audio with a content that the user can recognise immediately. For example, the image with
audio will be of the pastoral type if the sequence shows a landscape of fields or forests. For example, even the choice of the musical instruments may participate in the musical identification. Preferentially, the automatic music producing device is resorted to, of the type described in the PCT application published under number WO 00/17850, in the name of Rene-Louis Baron.
This device is based on a method which consists of:
- defining some musical moments in the course of which at least four notes are liable to be played,
- defining two families of note pitches for each musical moment, with the second family of note pitches having at least one note pitch which is not in the first family,
- setting up at least one sequence of notes having at least two notes, with each sequence being called a musical phrase, a sequence in which, for each moment, each note whose pitch belongs exclusively to the second family is surrounded exclusively by notes from the first family, and
- producing a signal representing each note pitch in each one of the said sequences. According to this method, two families of note pitches may also be defined for each musical moment, with the first family constituting a set of note pitches belonging to a chord duplicated from octave to octave. In this case, at the time of defining the two families of note pitches, the second family of note pitches may include at least the note pitches of a
range which are not in the first family of note pitches .
When at least one sequence of notes is formed having at least two notes, each musical phrase is, for example, defined as a set of notes for which the initial instants are not separated from one another, two by two, for more than a pre-determined duration.
In this known method, some parameters are envisaged made up by, especially, the definition of the musical moments, the definition of two families of note pitches and the constitution of at least one sequence of notes. According to the invention, preferentially, these parameters shall be made to depend on the digital signal representing the contents of the scene or sequence to be identified. For example, a war scene acts on a parameter thereby triggering off some drums and percussion instruments.
Whatever the mode for performing the automatic generation of the musical composition may be, in a preferred implementation the audio identification will also represent the duration of the pertinent video sequence .
Thus the invention concerns a method for file identification representing a process extending over a set duration, for example, a video sequence, which is such that an audio identification is associated with each file representing the contents of this file, with this audio identification having a short duration with respect to that of the file and being different from the latter.
Preferentially, the audio identification is musical and, to its advantage, produced with the aid of an automatic music generation device. In this case, the contents of each file may be identified by a digital signal, and this digital signal may be used to create some control parameters for the automatic music generation device.
The files are, in one embodiment, recorded on a medium and the audio identification is recorded on the same medium, under the form of control parameters for the automatic music generation device.
In a variation, the audio identification is recorded completely on the same media as the one containing the files. In one embodiment, the audio identification is also representative of the duration of the pertinent file. The audio identifications may be stored in a reserved zone in a medium for recording the files. Each audio identification may also be stored close to the zone, or in the zone, for recording the pertinent file. The invention is also related to a procedure for reading the files identified according to the procedure defined above in which the file identified is selected and the file selected in this way is consulted. The invention also concerns the application of the method defined above to the identification of video sequences from television broadcasts transmitted. It also relates to the application of this method to the identification of files, especially video sequences, on a disk, in particular those of the DVD type as well as to the identification of files on an audio tape,
especially in the form of a cassette. The invention also concerns a device for reading and/or recording audio identifications of files, especially video sequences, which includes an automatic music generation device.
The invention also concerns a device for reading file audio identifications which includes a device for selecting the file for which the identification is being listened to, and some means for consulting this file selected in this way.
Some other characteristics and advantages of the invention will appear with the description of some of its embodiments, with the latter being done by referring to the drawings attached herewith on which: Figure 1 represents a medium to which the procedure has been applied according to the invention,
Figure 2 is a scheme that is analogous to the one in Figure 1, for a variation,
Figure 3 is a scheme of a device for the reading of audio identifications in accordance with the invention, and
Figure 4 is a scheme showing one part of another device for reading audio identifications in accordance with the invention. In the example which will be described below, the labelling of the contents of the video scenes is referred to, which is standardised by the MPEG-7 standard. MPEG-7 files are used to describe multimedia documents at different levels, for example, the meaning and the contents, so as to optimise their indexing and their search in the Internet network and in databases.
An MPEG-7 file may be integrated in or attached to a document, or not yet form part of the document, but, in the latter case, it may then easily be found by means of a link. According to this standard, a label E± is associated to each scene content in range i, formed by a digital signal. The contents that may be labelled are, for example; a landscape, a first main character, a second main character, a secondary character, the contents of scenes such as a moving car, love scenes, scenes of violence, scenes from childhood, etc...
Each digital signal E is associated to some parameters of a procedure for automatic music generation of the type described in the document WO 00/17850, with these parameters representing some musical moments, some families of pitches for notes and at least one sequence of notes. These parameters are chosen in such a way that the music produced will identify the pertinent contents and, preferentially, the duration of the scene or sequence. The audio identifications produced this way are recorded on a medium 10 (Fig. 1) containing the set of video sequences, either in a reserved zone 12 of the table of contents type or close to (identification 16 - Figure 2) or in the same site as the video sequence 14. Whatever the site where the audio identification is located on the medium, this identification may be used to consult the different scenes or sequences of the video
involved; in this case, the audio identification or image replaces the fast display of the video sequences which, in general, is almost incomprehensible, whilst an image with audio is generally sufficent for the understanding, or impression as a whole, of the sequences .
Another use of the audio identifications, which mainly operates when the identifications are located in a reserved site, consists of providing direct access to the video sequence involved. In fact, in this case, the audio identification is associated to the site where the video sequence is located and a simple operation, such as pressing down an input key when the audio identification is heard, allows the user to access the start of the video sequence.
According to one variation (Fig. 3), which avoids recording the full audio identifications, each audio identification is represented only by the input parameters of the musical composition generating device, with such a generating device 18 being associated to the pertinent video player 20. This implementation thus allows the memory site to save space for the recording medium; it also allows an audio identification which presents some variations from one reading to another of the video sequences, which increases the interest of a search in the video sequence recording equipment. In fact, since the automatic music generation is based on random choices, the generation changes the input parameters from one reading to another, which allow audio identification whilst limiting the number of possibilities in a
restricted field but not involving the generation of a single type of music.
In the same film, different video sequences may present the same character. Thus, the number of different audio identifications is equal to the number of video sequences.
The audio identification, whether it is performed by direct recording or by the recording of the input parameters from the music composition generating unit, may be performed by the manufacturer of the video sequences, at the time of the editing or at the time of the post-production. It may also be performed at the time of viewing it later; the latter case applies when the medium on which the video is recorded allows for additional recording. For this purpose, in the example, the user has control keys to enter musical composition parameters, for example, on a remote control unit. Thus, the invention may be used not just for making films of the professional type but also for the making of films by private individuals using portable videocameras . It may be used for identifying films on video tape, on recordable disks or on any other support .
When the audio identification is recorded completely or only by its parameters- on a DVD disk, the different identifications are located, preferentially, in a reserved zone in the disk and, in this case, when the user listens to the audio identification, he/she may then, if he/she selects the pertinent sequence, directly access this sequence if the audio identification is associated with an
identification of the start of the sequence to be displayed.
When the recording medium is in the form of a tape, especially in the cassette type, the audio identification 22 (Fig 4) is located close to, preferably before, the video sequence identified 24. In this case, the reading device includes a buffer memory 26 in order to continue the reading of the audio identification as long as the zone for the video sequence has not been reached by the tape as it is played.
In one embodiment, all of the audio identifications have a duration of around a few percent, for example 3%, of the total duration of the video recording. Thus the music recording for a video film with a duration of around one and a half hours will have a duration of around three minutes, which corresponds to the duration of a song.
As stated above, the procedure for identification from the invention allows sequences to be selected, to be compiled and then to view them.