DE69521597T2 - Device with microprocessor for the selection and recognition of pieces of music - Google Patents

Description

The present invention relates to a device with a microprocessor for selecting and recognizing pieces of music.

In the field of music, and in particular in the field of electronic musical instruments, the user very often needs to recall musical pieces for study purposes. To do this, he must search for the piece that has already been stored and recall it.

Normally, all electronic musical instruments are provided with a special archive in which a number of pieces are stored. The user can retrieve a piece using a function specifically designed to search and play, so that the user can listen to the piece and use it for whatever purpose he or she wishes.

This process obviously takes time and can lead to errors on the part of the user, where two or more pieces may be laid out similarly and have many notes in common. This requires a series of decisions on the part of the user, which require attention and concentration, so the user tries to carry out such work only when he has sufficient time and peace available for this purpose.

From EP-A-0372 a device is known today which selects pieces of music, but not recognises the pieces mentioned.

According to the document US A 5250745, the selection of music pieces is carried out by adapting to the type of voice in terms of sound and treble extension, but the recognition of the mentioned pieces does not take place.

Therefore, there is currently no automatic device available for the quick and reliable selection and recognition of a piece of music requested by the user.

The present invention aims to avoid the technical problems described above by providing a device with a microprocessor as in claim 1 for selecting and recognizing musical pieces, whereby a musical tone is recognized by comparing a performance provided by a source (such as a microphone, MIDI input or keyboard) with musical tones stored in an archive. Regardless of the musical ability of the user, the device can automatically and quickly recognize and deliver a musical piece on the basis of a minimum of notes that make it up, regardless of the octaves and the key to which the piece is played.

The above purpose and others will be better clarified hereinafter and are achieved by a microprocessor-based device for selecting and recognizing musical pieces as defined in the appended claims.

Further features and advantages will become apparent from the description of a preferred embodiment of the invention, which is given below under Reference is made to a single figure, given purely as an illustrative and non-limiting example, showing a schematic block diagram of the device according to the invention.

In Fig. 1, 1 indicates a microphone used to record musical sources playing a piece. A pitch-to-MIDI type sound/note converter 2 connected to the microphone 1 generates musical notes. A MIDI interface 3 can be connected to any musical instrument 11, while 4 indicates a keyboard on which any type of musical piece can be played.

All of the above-mentioned means for generating musical pieces 2, 3 and 4, which can be used together or independently of each other, send their musical sequences to means for executing musical pieces, designated in the figure by a block 10, which essentially consist of a sound generator, amplifiers and stereo systems.

Between the means for executing pieces of music 10 and the means for generating 2, 3 and 4 pieces of music, a control console 5 is arranged, which control console 5 has, among other things, a switch 5a that allows a redirection from a first contact A to a second contact B.

The control console 5 is managed by a microprocessor 6 of a type always present in every electronic musical instrument or sound reproduction element. The microprocessor 6 determines, through its own activation means (not shown here), a switching from the said first contact A to said second contact B.

A recognition microprocessor 7 compares the notes received from the generating means 2, 3 and 4 with libraries 9 which contain a number of archives in encrypted form relating to the pieces of music.

A display 8 shows the title of the piece and the user can also observe the ongoing selection procedure.

What follows is a functional description of the device.

The generating means 2, 3 and 4 are used to generate a sequence of musical notes which are then compared with the pieces present in the libraries 9 of the microprocessor 7 for recognition in order to check whether the piece requested by the user is contained therein.

Thus, the recognition microprocessor 7, together with the libraries 9, constitutes a logical discrimination means by which a logical comparison is made between the digitized musical notes received from the generating means 2, 3 and 4 and those present in the libraries 9, so that a preselection and a recognition in real time of the stored musical pieces containing the notes sent by the generating means 2, 3 and 4 can be carried out.

All these processes take place when the switch 5a is moved from the first contact A to the second contact B so that the corresponding procedure can be activated.

The means of logical differentiation mentioned above provide for a sequential Number of process flows, which are divided as follows.

Each piece of music is characterized by a recognition pattern, which is an encryption of a sequence of notes that identifies the specific piece.

By means of the recognition microprocessor 7, a search is carried out in the libraries 9 for a piece that has the same recognition pattern as the input sequence. If a recognition pattern matches, the piece of music is considered to be recognized and is sent to the means for executing pieces of music 10, which consist of a musical instrument to be played. If, however, the recognition pattern does not match exactly, the microprocessor 7 identifies, with the aid of an evaluation program, the piece in the libraries that is most similar to the input piece. When a piece has been recognized, its title appears on the display 8.

If recognition or near-recognition has been made, the user can decide at this point whether he wants to play through the piece using the control console 5.

The invention contains a particularly original solution for recognizing music pieces, regardless of the octaves and keys according to which they are entered.

Each note is provided with a MIDI number key by the microprocessor 7 for recognition; said number is fixed and normalized, being considered as the remainder of the MIDI number key after division by twelve (twelve being the number of notes in a musical octave).

For example, if the MIDI key for a note is 56, the value assigned to that note is 8, which is the remainder of dividing 56 by 12.

Then, in order to detect when the entered notes have been read in order to find the desired piece of music, the microprocessor 7 calculates the algebraic difference between the "normalized" values of one note and the following note. Each element calculated in this way can take values ranging from minus 11 to plus 11, since each note has normalized values from 0 to 11.

Essentially, the recognition pattern is organized according to a two-way arrangement, as explained below: patt-recogn (max lib songs) (max recogn notes), where max lib songs means the maximum number of pieces contained in the libraries 9, which in the case of the applicant is 512.

Max recogn notes is the maximum number of notes used to recognize a piece being searched for. In the applicant's case, the number is eight. Of course, the numbers can be changed both by max lib songs and by max recogn notes.

Each element of the array is represented by a byte, which, as stated above, has a value range between -11 and +11. The recognition patterns thus consist of a numerical sequence of values that represent the difference between one note and the next.

The actual setting is done using a program that is based on two one-way arrangements, as follows:

- search (max lib songs) and errors (max lib songs).

In the first unidirectional arrangement, all pieces have a corresponding recognition pattern which is equal to the stored input patterns; the patterns produced by the generating means 2, 3 and 4 are compared two at a time. This first arrangement is designated by a word, i.e. by 8 bits.

An operation known as "read inputs" is provided in the search array. This operation contains three sub-operations, each of which has the purpose of analyzing the incoming notes coming from the sources, i.e. keyboard 4, interface 3 and note/note converter 2.

Once the difference between the first two notes of the piece sent by the generating means 2, 3 and 4 has been calculated, the recognition microprocessor 7 begins to scan all the pieces present in the libraries 9 and stores all those that correspond to the piece entered: this is carried out by a sub-process of the initial search.

This comparison is then repeated with the help of a next search sub-process for all the other notes sent by the generating means 2, 3 and 4. The comparison process ends when the search arrangement recognises the piece, namely with an exit recogn sub-process in which song status - song recognised is described and thus the state of identification is determined.

Of course, this happens when the piece is recognized, that is, when the search order confirms a perfect match between a piece contained in max lib songs and the sequence of notes sent by the generating means 2, 3 and 4.

If errors are indicated, which is indicated by an 8-bit arrangement errors occurs, a clear errors subroutine is triggered and an internal search is performed in the libraries 9 to determine the difference between the samples of the submitted notes and the corresponding note sample of at least one piece contained in max lib songs.

The above-mentioned search is carried out within the sub-process calc errors. In addition, since an error threshold has been set beyond which no detection takes place, in this case a sub-process called min errors is activated, which allows the most similar piece to be detected and thus the identification of a song status - min errors.

In the applicant's implementation, this error threshold was given numerically as 6; this is purely a reasonable value, determined on the basis of tests carried out.

If there are more with the same minimum value in terms of errors, the procedure always selects the last selected piece. In this case, it results in a song status - song selected.

If a song status - sond recognised result is obtained, this means that a piece has been perfectly recognised; if, on the other hand, a song status - min errors appears, a piece that appears most similar has been recognised within the limits of the error threshold. If a song status - not recognised result is obtained, no piece has been recognised in libraries 9 that have note patterns corresponding to those entered by generation means 2, 3 and 4, i.e. there is an error range that exceeds the specified error threshold.

In summary, the logical discrimination means represented by the sequential operations of the combination of the recognition microprocessor 7 and the libraries 9 allow the emission, via the switch 5a, of a series of notes represented by a musical passage and executed by one of the generation means 2, 3 and 4, with the aim of bringing about a comparison between said libraries 9 according to a sequence of procedures providing for the analysis of two notes simultaneously, of course digitized, coming from said generation means 2, 3 and 4, and a comparison between the recognition patterns of said notes, represented for example as a sequence of encrypted numbers and recognized by the difference between an encrypted note and the next note, and a recognition pattern represented by the corresponding notes of the stored musical pieces. The said methods also comprise reading the notes, selectively searching to identify said recognition patterns and continuing said search until a single identification is made among the patterns in the search order. If this match is exact, a message song status recognised is sent, i.e. the recognised piece can then be played by the instruments 10; if this recognition is not received, a message song status not recognised is sent and the piece is not played. If more than one pattern is recognised, a message song status selected is received which provides for a min error, i.e. the recognition of a piece with a minimum number of errors according to a specified tolerance in relation to the reference piece.

The method comprises a sequential search which begins with a flow called first search, continues with another search called sub-flow, and ends with a sub-flow called exit recogn, with which the program managing all the above-mentioned phases identifies either an error condition or a detection condition.

With the above sequential procedure it is possible to identify from a series of input notes, in whatever tone and in whatever octave they may be, those pieces which have the same content as the corresponding numerically encrypted notes in the libraries 9 in which a number of reference pieces are stored.

The piece you are looking for can now be played. This procedure provides a well-adjusted margin for error, speeds up the selection and recognition, and avoids errors by the user.

Claims (8)

1. Device with a microprocessor for selecting and recognizing musical pieces, characterized in that it comprises means (1, 2, 3, 4) for generating digitized information representing musical notes performed by a user, which are supplied in encrypted form to the device in order to access a number of libraries (9) contained in a memory , which libraries (9) consist of musical reference notes for musical pieces; said generating means and said memories being managed by a first microprocessor (7) to which said memories are connected and by a second microprocessor (6) to which said generating means (1, 2, 3, 4) are connected, said first microprocessor (7) also managing means for logical discrimination by comparing input notes by comparing the notes sent by said generating means via a corresponding enable switch (5a) with said libraries (9) in order to select and recognize in real time pieces of music containing said notes supplied to the device. 2. Device with microprocessor according to claim 1, characterized in that said means for logical differentiation (7) carry out sequential processes, which define recognition patterns and include means for carrying out a first sequence phase in which the input notes are read in encrypted form; means for carrying out a search phase in which two of the input notes are read simultaneously; and comparison means which compare an input sequence with a stored sequence in the libraries (9), the comparison between said input and stored sequences being represented as a numerical sequence showing a difference between an encrypted note in numerical form having a value from 0 to 11 and a following similarly encrypted note, and means for deciding that, if there is a perfect match, a piece of music has been recognized. 3. Device according to claim 2, characterized in that a recognition error threshold is provided if no perfect recognition of the piece of music has taken place, below which recognition error threshold a piece of music can be identified as such that contains a minimum error in relation to the notes entered. 4. Device according to claim 2, characterized in that said means for logical discrimination (7) contain means for representing a situation in which the piece of music to be recognized is not identified if there is no equality between a sequence of numerical, encrypted notes stored in the memory (9) and a numerical, encrypted sequence that corresponds to the entered sequence of notes. 5. Device according to claim 2, characterized in that said means for distinguishing (7) provide a numerical time-direction arrangement according to which said musical reference notes are organized. 6. Device according to claim 4, characterized in that the said recognition pattern provides two unidirectional arrangements, one of which relates to the search for pieces of music, while the other relates to any recognition error. 7. Device according to claim 6, characterized in that each bit of each arrangement takes a value between -11 and +11 . 8. Device according to claim 2, characterized in that the said numerical sequence, which represents a series of differences between encrypted notes, is calculated according to a MIDI standard key.