JP2002132278A - Method for retrieving song through melody signal - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method which enables users to retrieve songs through melody signals and to retrieve the songs even when the users forget or mistake or do not know text, music names or singer names. SOLUTION: This method enables the users to retrieve the necessary songs from a database for the songs by humming, singing or knocking the syllabism of the melodies of the songs or playing the same with a musical instrument. The melodies inputted by the unrest are digitally processed and are converted into an intermediate format which is then compared with the intermediate song format from the database of the songs. The users are able to find the necessary songs.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for searching for a song through a melody signal, and more particularly, to a method for a user to hum or sing a segment of a song.
They may knock or play with an instrument, and the incoming melody signal may be digitally processed and converted to an intermediate format, which may be the intermediate format for the song in the song database. Compared, thereby allowing the user to retrieve the required song from a song database.

[0002]

In the prior art, text such as lyrics, song names, or singer names are used by input actions and through comparisons, and the requested song is retrieved from a song database. However, the above technique is ineffective if a user who only has an impression of the melody or rhythm of the song forgets, mistakes, or does not know the general conditions of the lyrics, song title, or singer name. Will be.

Accordingly, the present invention seeks to avoid the inconvenience of the prior art by searching for a song through a melody signal and allowing the user to forget the lyric, song title, or singer's name, to make a mistake or not to know it. Provide a way to search for.

[0004]

SUMMARY OF THE INVENTION It is an object of the present invention to provide a method for searching for a song through a melody signal, so that the user can forget, wrong or wrong lyrics, song titles, singer names, or the like. Even if you do not know, it is possible to search for a required song only by impression of the melody or rhythm of the song.

[0005]

SUMMARY OF THE INVENTION The method described above comprises a procedure A for acquiring a melody signal, i.e., a melody or rhythm input through an electronic peripheral device such as a microphone or other musical instrument. Procedure B, which memorizes the melody signal digitally, ie, the melody or rhythm collected in Procedure A, uses digital signal processing (hereinafter referred to as "DSP") to store the melody signal digitally. Complete the process of analysis (or called pitch tracking)
Thereafter, a procedure in which those fundamental frequency signals are converted into a specific intermediate format to prepare for a procedure C for melody comparison, and a procedure C for comparing melody of a song, that is,
Musical signals, such as frequency and rhythm, are converted to an intermediate song format, after which the intermediate song format in the song database is compared to the particular intermediate format converted by procedure B, The procedure in which the similarity between the two intermediate formats is detected, and the procedure D, that is, the song in the database of songs most similar to the song requested by the user, is performed in the above intermediate procedure detected by the procedure C. That the format is detected based on the similarity between the formats.

[0006]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Please refer to both FIG. 1 and FIG. The present invention retrieves a song through a melody signal and retrieves the required song from the song database by the user humming, singing, knocking, or playing with a musical instrument the song segment. Provide a way to do that. This method includes procedures A to D
Is included.

Procedure A collects the melody signal. In this procedure, a user enters a melody or rhythm of a song through an electronic peripheral. Enter the rhythm of the song, WAVE, R
A user can hum, sing, or knock a few verses of a song through a microphone in order to edit and store it in a particular electronic data format, which is a digital sound format such as A or AU. Various input methods are used, including playing a melody or rhythm with a musical instrument, or pre-editing a melody or rhythm by using an electronic device such as using a computer keyboard. It is possible. The user, through the recording device,
The melody signal resulting from humming, singing, or knocking the melody of the song can be converted to the same electronic data format and then stored in the storage of the electronic device.

[0008] Procedure B processes the electronic digital data format collected in procedure A by performing fundamental frequency analysis (also called pitch tracking) through a DSP. The fundamental frequency for each segment of the signal is detected and converted to a corresponding note or quantified value. Finally, the above fundamental frequency is
Further, it is converted to an appropriate intermediate format for the melody comparison in the procedure C.

[0009] The above fundamental frequency analysis (pitch tracking) is based on the digital sound signal obtained from the procedure A.
The format is converted through a DSP, such as a Fourier transform, a fast Fourier transform, or a wavelet transform, into a variance in the frequency domain where the high and low frequency parts are separated, so that the fundamental frequency for each segment of the signal is detected. Process. The above intermediate formats are Ternary Contour,
A digital sound format that includes the frequency or rhythm of a song, such as Numerical Contour or Chord.

[0010] Ternary contours use three symbols U, D, and S to represent the relationship between one note and another.
Is an intermediate format used. U represents a note that is higher than the previous note, D represents a note that is lower than the previous note, and S represents the same note as the previous note. Thus, the notes do, les, mi, mi, and les can be denoted as U, D, S, and D.

[0011] Numerical contour is an intermediate format in which numerical values are used to represent the relationship between one note and another by using semitones as a metric. 1 represents a note one semitone higher than the previous note, -1 represents a note one semitone below the previous note, and 0 represents the same note as the previous note. Thus, when using note C as the base note, note C, re, mi, mi, and re can be denoted as 2, 2, 0, and -2.

A chord is an intermediate format in which a set of notes is represented by being converted to its corresponding chord. Thus, the notes C, D, M, M, and M can be denoted as C, Dm, Em, Em, and Dm.

Procedure C compares song melodies. In this procedure, musical signals, such as frequency and rhythm, are converted into a specific intermediate song format for the song as a whole. Thereafter, the intermediate song format in the song database will be compared to the particular intermediate format converted by Procedure B. During that time, comparison means such as dynamic programming, dynamic time warping, artificial neural networks, and fuzzy logic are available. For example, the intermediate format of Ternary Contour can be used to compare through a word string. The purpose of Procedure C is to use two forms, both from the intermediate format created in Procedure B from the input melody signal and from the intermediate format created from the song database in Procedure C.
Is to find similarities between two intermediate formats.

The procedure D is the output of the result. In this procedure, the similarity between the intermediate formats is determined by the procedure C.
A song in the song database most similar to the song requested by the user is found based on what is found by the user.

Finally, in a preferred embodiment of the present invention, a suitable interface device, such as a computer monitor, is used to display a list of songs in a similarity order so that the user can select them. .

Although the present invention has been described in considerable detail with reference to certain preferred embodiments, those skilled in the art will recognize that various modifications may be made within the spirit and scope of the appended claims. Will be readily apparent.
Therefore, the spirit and scope of the appended claims should not be limited to the description of the preferred embodiments herein.

[Brief description of the drawings]

FIG. 1 is a schematic diagram illustrating the method of the present invention for searching for a song through a melody signal.

FIG. 2 is a flow chart of a method for searching for a song through a melody signal according to a preferred embodiment of the present invention.

Continuing on the front page (72) Inventor Chang Chi Singh Taiwan, Xin Chu, Euan Hou Street, number 22, 3F F-term (reference) 5B075 ND14 NK06 PP28 PR06 QM08 5D108 BA06 BA32 BC08 BC17 BC20 BD08 5D378 KK44 MM34 MM52 MM63 MM97 XX43

Claims (7)

[Claims] A method for retrieving a song through a melody signal to enable a user to retrieve a song from a song database by using the melody signal, the system comprising: A melody or rhythm of any segment of a song input to the melody or rhythm into a data format readable by the system; and a melody or rhythm collected in the melody or signal acquisition procedure is converted into a digital signal. Digitally processing the melody signal, if various melodies are found and converted to corresponding notes or quantified values, and further converted to an equivalent intermediate format, The intermediate format converted through the digital processing procedure of And intermediate format comparison procedure to be compared with intermediate song formats from a song in the scan, and results output procedures song is output according to the degree similar for the comparison result from the intermediate format comparison procedure,
A method that includes 2. The system according to claim 1, wherein the system is a personal computer, an Internet server computer, a karaoke machine, a jukebox, or another electronic device having a function of processing a melody signal. The described method. 3. The method according to claim 1, wherein the peripheral input device is a microphone, a keyboard, a musical instrument, a recording device, or another input generating system capable of reading a melody signal. 4. The melody signal includes humming by a user,
The method of claim 1, wherein the method is a singing, knocking, or playing a musical instrument, a recorded melody signal, or a digital melody signal. 5. The method of claim 1, wherein the data format readable by the system is a WAVE, RA, AU, or other digital melody signal. 6. The melody analysis according to claim 1, wherein the melody analysis is a fundamental frequency analysis, a ternary contour, a numerical contour, a chord, or another means having a function of performing the melody analysis. The method described in. 7. The method of claim 1, wherein the comparing comprises: dynamic programming, dynamic time warping, artificial neural network,
The method according to claim 1, wherein the method is fuzzy logic or other means having a function of comparing intermediate formats.