CN112201100A - Music singing scoring system and method for evaluating artistic quality of primary and secondary schools - Google Patents

Music singing scoring system and method for evaluating artistic quality of primary and secondary schools Download PDF

Info

Publication number
CN112201100A
CN112201100A CN202011162757.2A CN202011162757A CN112201100A CN 112201100 A CN112201100 A CN 112201100A CN 202011162757 A CN202011162757 A CN 202011162757A CN 112201100 A CN112201100 A CN 112201100A
Authority
CN
China
Prior art keywords
rhythm
scoring
note
characteristic sequence
score
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202011162757.2A
Other languages
Chinese (zh)
Inventor
雷小林
胡健
蒋文颉
张震
郑婧
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jinan University
Original Assignee
Jinan University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Jinan University filed Critical Jinan University
Priority to CN202011162757.2A priority Critical patent/CN112201100A/en
Publication of CN112201100A publication Critical patent/CN112201100A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
    • G09B—EDUCATIONAL OR DEMONSTRATION APPLIANCES; APPLIANCES FOR TEACHING, OR COMMUNICATING WITH, THE BLIND, DEAF OR MUTE; MODELS; PLANETARIA; GLOBES; MAPS; DIAGRAMS
    • G09B7/00—Electrically-operated teaching apparatus or devices working with questions and answers
    • G—PHYSICS
    • G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
    • G09B—EDUCATIONAL OR DEMONSTRATION APPLIANCES; APPLIANCES FOR TEACHING, OR COMMUNICATING WITH, THE BLIND, DEAF OR MUTE; MODELS; PLANETARIA; GLOBES; MAPS; DIAGRAMS
    • G09B15/00—Teaching music

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Educational Administration (AREA)
  • Educational Technology (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Reverberation, Karaoke And Other Acoustics (AREA)

Abstract

The invention discloses a system and a method for scoring music singing in the evaluation of the quality of art in middle and primary schools, wherein the system comprises a server, a firewall, a communication network, a computer and a mobile terminal; the server is a server cluster formed by a plurality of physical devices; the communication network is a wired network or a wireless network and is mainly used for connecting a server, a computer and a mobile terminal to perform data interaction; the method can identify the note starting point in the audio frequency of the singer, calculate the rhythm duration and pitch characteristic sequence, match and compare the identified characteristic sequence with the reference standard scoring characteristic sequence, and perform scoring result and scoring detail according to the comparison result, thereby improving the scoring accuracy, well meeting the requirement of objective and accurate scoring of the note singing in the quality of art evaluation of the primary and secondary school students, providing instructive suggestions for improving the singing level of the singer, and enabling the student to carry out targeted correction and strengthening exercises on the point of losing score.

Description

Music singing scoring system and method for evaluating artistic quality of primary and secondary schools
Technical Field
The invention relates to the technical field of audio recognition, in particular to a system and a method for evaluating Chinese music singing and scoring in the process of evaluating the quality of art in primary and secondary schools.
Background
At present, the education department performs art evaluation in primary and secondary schools in China, the music art evaluation is one of evaluation subjects, and the music art detection also becomes an important daily teaching work of a music teacher in the primary and secondary schools. Traditional music singing scoring techniques and means generally rely on human ear hearing to score, which is highly dependent on human subjective judgment, and different scoring people may give different scoring results. Meanwhile, manually listening to a large amount of singing audio is a work which consumes a large amount of time and energy, so that the method is difficult to be applied to large-scale batch evaluation work.
The existing karaoke scoring system can give scores to singing results, but the scoring is not strict and accurate enough in interest property, and music art evaluation needs to give objective and accurate scoring results to the singing results by referring to a scoring standard and give corresponding scoring and losing details as much as possible. Therefore, the existing karaoke scoring system is not suitable for scoring music singing.
The existing humming recognition system mainly compares the characteristic sequence of the audio with the characteristic sequences of songs stored in a song library to find out the song which is most matched with the characteristic sequence in the song library, focuses on the song similarity matching and searching functions, is mainly used for music classification or music recommendation scenes, and is not suitable for quantitatively scoring a singing result by a similarity matching algorithm and cannot give a comparison result of each beat in the audio.
Disclosure of Invention
The invention aims to provide a system and a method for scoring a music singing in the process of evaluating the artistic quality of primary and secondary schools, which aim to solve the problems in the background technology.
In order to achieve the purpose, the invention provides the following technical scheme: a music singing scoring system for middle and primary school artistic quality evaluation comprises a server, a firewall, a communication network, a computer and a mobile terminal;
the server is a server cluster formed by a plurality of physical devices;
the communication network is a wired network or a wireless network and is mainly used for connecting the server, the computer and the mobile terminal to perform data interaction.
A method for evaluating the vocal singing score of the middle and primary school artistic quality comprises the following steps:
step 1: the administrator inputs the standard characteristic sequence of the reference test questions through the computer and uploads and saves the standard characteristic sequence to the server; the reference test questions are mainly divided into two types: rhythm type examination questions and melody type examination questions, wherein the rhythm type examination questions generally only examine the mastering conditions of singers on music rhythm, the standard characteristic sequence and the scoring characteristic sequence of the rhythm type examination questions are music rhythm codes, the melody type examination questions generally simultaneously examine the mastering conditions of singers on music rhythm and music melody, and the standard characteristic sequence and the scoring characteristic sequence of the rhythm type examination questions are music rhythm codes and music pitch codes;
step 2: a singer inputs and uploads an audio file through a mobile terminal such as a smart phone, the singer checks the type of a test question, and if the test question is a rhythm test question, the singer only needs to sing a music rhythm towards a mobile phone microphone; if the test question is melody type test question, the mobile phone microphone needs to be sung music melody, including rhythm and pitch;
and step 3: the server acquires a scoring feature sequence of the audio file to be scored through a feature extraction algorithm, and matches the scoring feature sequence with a standard feature sequence; the rhythm type examination questions only need to be matched with the rhythm characteristic sequence, and the rhythm type examination questions need to be matched with the rhythm characteristic sequence and the pitch characteristic sequence;
and 4, step 4: and 3, calculating a score according to the matching result of the step 3 and a scoring rule, and returning the score and scoring details.
Further, according to the data of the reference standard test questions input in the step 1, the type of the reference standard test questions is input; inputting rhythm examination type test questions into rhythm duration codes of each bar of standard test questions; the rhythm examination type test question is recorded with rhythm duration code and pitch code of each bar of the standard test question.
Further, according to the step 2, the singing audio of the singer is converted into the wav standard format through the audio format.
Further, noise reduction processing is carried out on wav standard audio of the singer, and feature extraction is carried out according to the type of the singing test questions. And extracting rhythm characteristics of the rhythm examination type test questions through an audio rhythm characteristic extraction module according to the sound amplitude change condition, and calculating to obtain a rhythm characteristic sequence. The rhythm examination type test question completes note segmentation through the note segmentation prediction analysis module, the duration of each note after segmentation is calculated, the rhythm feature sequence of the whole audio is obtained, and the pitch feature extraction module extracts pitch features of each note after segmentation and obtains the pitch feature sequence of the whole audio.
Furthermore, the obtained rhythm characteristic sequence, pitch characteristic sequence and standard score characteristic sequence are matched through a characteristic sequence matching module, and the matching result, the number and the position of the multi-singing notes and the number and the position of the missed-singing notes of each note in the standard score characteristic sequence are found out.
Furthermore, the score and result display module calculates the score of each note according to the set score weight, the total score and the standard score characteristic sequence number, further obtains the score of each note according to the matching result of each note, deducts the related score according to the number of the multi-note and the number of the missed note, subtracts the deducted score from the score summation result of all notes to obtain the final score, and simultaneously displays the score details according to the matching result of each note, the number of the multi-note and the number of the missed note.
Further, the audio is converted into a spectrogram through a CQT algorithm, original data samples are cut according to a fixed window width by taking the wave trough of the envelope as the center according to a time sequence in combination with a music amplitude envelope wave trough detection algorithm to form small segments of the spectrogram, and positive and negative samples are labeled in a mode of manually judging whether the small segments of the spectrogram contain note starting points. And training positive and negative samples by adopting deep learning, and storing the trained network model.
Compared with the prior art, the invention has the beneficial effects that: the voice singing evaluation method has the advantages that note starting points in audio of singers can be identified, rhythm duration and pitch characteristic sequences are calculated, the identified characteristic sequences are matched and compared with reference standard scoring characteristic sequences, scoring results and scoring details are carried out according to comparison results, scoring accuracy is improved, the requirement that the singing scoring of the music in the quality of art test and evaluation of primary and secondary school students is objective and accurate is well met, instructional suggestions are given for the singers to improve singing levels, and students can carry out targeted correction and strengthening exercises on the points of losing scores.
Drawings
FIG. 1 is a schematic illustration of an implementation environment to which various embodiments of the present invention relate;
FIG. 2 is a block diagram of the music performance scoring system of the present invention;
FIG. 3 is a schematic diagram of a rhythm examination type test question code of the present invention;
FIG. 4 is a schematic diagram of a melody examination type test question code according to the present invention;
FIG. 5 is a flow chart of rhythm examination type audio rhythm feature extraction of the present invention;
FIG. 6 is a flowchart of the melody examination type audio rhythm and pitch feature extraction of the present invention;
FIG. 7 is a flowchart of the melody examination type audio note onset segmentation based on deep learning according to the present invention;
fig. 8 is a diagram of a rhythm-examined type audio rhythm recognition result of the present invention;
fig. 9 is a diagram of the rhythm examination type audio rhythm and pitch recognition result of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention provides a technical scheme that: a system and a method for scoring music singing in the evaluation of the artistic quality of primary and middle schools are provided.
FIG. 1 is a schematic diagram of a system implementation environment of the present invention, including: the system comprises a server, a firewall, a communication network, a computer and a mobile terminal;
the server may be an independent single physical device, a server cluster formed by a plurality of physical devices, or a cloud computing center. The algorithm and the software service related to the invention are mainly provided, and are the running environment of a background system;
the computer can be a desktop computer and is mainly used for inputting relevant information of examination standard test questions, acquiring and uploading test audio and displaying scoring results;
the mobile terminal can be a smart phone, a tablet personal computer, a portable computer and the like and is mainly used for collecting and uploading test audio and displaying scoring results;
the communication network can be a wired network or a wireless network, and is mainly used for connecting a server, a computer and a mobile terminal to perform data interaction.
Fig. 2 is a flow chart of the audio scoring method of the present invention, comprising the steps of:
step 1, a manager inputs a standard characteristic sequence of a reference test question through a desktop computer, and uploads and stores the standard characteristic sequence to a server; the reference test questions are mainly divided into two types: rhythm type examination questions and melody type examination questions, wherein the rhythm type examination questions generally only examine the mastering conditions of singers on music rhythm, the standard characteristic sequence and the scoring characteristic sequence of the rhythm type examination questions are music rhythm codes, the melody type examination questions generally simultaneously examine the mastering conditions of singers on music rhythm and music melody, and the standard characteristic sequence and the scoring characteristic sequence of the rhythm type examination questions are music rhythm codes and music pitch codes; the rhythm code is shown in fig. 3, and the pitch code is shown in fig. 4;
step 2, a singer inputs and uploads an audio file through a mobile terminal such as a smart phone, the singer checks the type of the test question, and if the test question is a rhythm test question, the singer only needs to sing a music rhythm towards a mobile phone microphone; if the test question is melody type test question, the mobile phone microphone needs to be sung music melody, including rhythm and pitch;
step 3, the server acquires a scoring feature sequence of the audio file to be scored through a feature extraction algorithm, and matches the scoring feature sequence with a standard feature sequence; the rhythm type examination questions only need to be matched with the rhythm characteristic sequence, and the rhythm type examination questions need to be matched with the rhythm characteristic sequence and the pitch characteristic sequence;
and 4, calculating scores according to the matching results in the step 3 and scoring rules, and returning scores and scoring details.
Fig. 5 is a flow chart of rhythm examination type audio rhythm feature extraction and scoring of the present invention, which mainly relates to a rhythm examination type audio rhythm extraction and scoring method, comprising the following steps:
step 11, inputting subject information and a standard rhythm characteristic sequence of rhythm examination questions, mainly rhythm codes, and referring to step S1 in the specific implementation process.
And step 12, the singer inputs and uploads the audio file through a mobile terminal such as a smart phone. The singer checks the test question information, sings the music rhythm towards the mobile phone microphone, and submits and uploads the singing audio. The background server 140 performs noise reduction preprocessing on the uploaded audio to be scored, and then obtains an audio amplitude oscillogram.
Step 13, the rhythm segmentation method based on the amplitude difference threshold value of the reference feature prior information cuts out note starting points according to the amplitude period significance variation characteristics, calculates the duration of each note according to the interval of the note starting points, and confirms the rhythm type of each note based on the statistical characteristics of the note duration and the continuity of the preceding note and the following note to obtain the rhythm feature sequence of the whole audio. The method comprises the following specific steps:
step 1301, after the noise reduction processing is performed on the audio signal, the short-time root mean square energy of the information is used as the envelope of the signal amplitude to obtain the envelope curve of the whole audio, wherein the short-time root mean square energy is
Figure BDA0002744879080000061
In the formula, a2(N) is the square of the amplitude of the ith frame of windowed music signal, and N is the window length.
Amplitude difference function of DE(n)＝En+1-En
Step 1302, smoothing the envelope curve in step 1 by adopting a median filtering mode, and removing burrs on the envelope curve caused by environmental noise, sound ray instability and the like;
step 1303, extracting all peak values in the envelope curve according to a local maximum method, including position information of the peak values, sorting the peak values according to magnitude, and taking an average value of N peak values with the largest peak value as an initial threshold value
δ＝avg(Top(Pn，N))
In the formula, Pn is a peak sequence, and Top represents the maximum N values in the peak sequence.
Step 1304, multiplying the initial threshold obtained in step S1303 by a segmentation coefficient γ (usually γ is a percentage of the initial threshold, and the reference value is 40% according to the statistical characteristics of the collected sample data), calculating a rhythm starting position determination threshold, and adding all determined rhythm starting positions to a rhythm starting position set PSIn (1).
Etheshold＝δγ＝avg(Top(Pn，N))γ
The rhythm starting position is determined according to the criterion
Figure BDA0002744879080000062
Wherein p is the sequence number of the audio frame
Step 1305, confirming an initial point and an end point of the non-mute section of the audio, calculating the duration of the audio, and calculating the reference duration of each rhythm type by combining the rhythm duration codes of the reference standard test question:
Durationn＝T×Rs/4000 Rs∈(R1、R2、...、Rn)
in the formula of RhythmsThe numbers of different rhythms are 125, 250, 500, 1000 and the like.
Step 1306, based on all the rhythm start positions P calculated in step 4SEach time of calculationDuration T of a rhythmi＝Ps+1-PsAnd confirming the alternative rhythm type according to the rhythm duration, wherein the judgment rule is as follows:
Figure BDA0002744879080000071
wherein abs () is absolute value operation and N is set of all rhythm types of reference standard test question
Step 1307, converting all the alternative rhythm types of the reference standard feature sequence and the audio into corresponding recognition character sequences. The reference conversion relationship is as follows:
rhythm name Code number Corresponding character Rhythm name Code number Corresponding character Rhythm name Code number Corresponding character
Full rhythm 4000 A Is divided into two parts 2000 B Point of attachment is halved 3000 C
Four-section 1000 D Four points with points 1500 E Eight minutes 0500 F
Eight points on the table 0750 E Sixteenmo 0250 G Sixteenth minute from the point of attachment 0375 H
Thirty-two parts 0125 J
Step 1307, finding out the matching result of the two character sequences by using a fast character string matching algorithm, and if the standard character sequence has unmatched characters, obtaining an alternative rhythm starting point set P by referring to the method of step 4 and using a lower rhythm starting position judgment threshold valueoThe rhythm starting position p corresponding to the previous character and the next character of the characterfAnd paFrom a set of alternative tempo starting points PoChecking whether a missing starting point exists or not, and judging the conditions as follows:
Figure BDA0002744879080000081
Figure BDA0002744879080000082
in the formula TfIndicating the duration of the latter rhythm, TaIndicating the duration of the latter rhythm, TcIndicating the duration of the current tempo.
DurationfIndicating that the latter tempo matches the reference Duration, of the standard tempoaDuration, representing the matching of the latter tempo to the standard tempocIndicating the duration of the current tempo.
If the judgment conditions are all satisfied, the starting position p of the current alternative rhythm is determinedcJoining a set PSIn (1). If not, skipping.
And step 14, performing character string fast matching on the rhythm characteristic sequence and the standard rhythm characteristic sequence of the reference test question, and finding out the number of multi-singing rhythms, the number of missed singing rhythms and a matching alignment result.
And step 15, obtaining the score of each note, calculating a final score according to the matching alignment result, the number of multi-singing rhythms and the number of missed singing rhythms and a scoring rule, and returning the final score and scoring details.
The character string fast matching criterion used in step 14 includes the following steps:
1401, judging the number of the reference characteristic sequence and the scoring characteristic sequence, if the number of the reference characteristic sequence and the scoring characteristic sequence are equal, comparing the characters one by one, and calculating the matched character sub-string ScerrectPosition P of matched charactercerrectPosition P of characters on unmatchedmissAnd is formed by Scerrect、PcerrectAnd PmissCalculating hit number NcorrectThe number of missed singing NmissSince the number of the reference characteristic sequence and the scoring characteristic sequence is equal, the rhythm of multi-singing does not exist, so the number of the multi-singing N is equalextra0, wherein Ncorrect+NmissTotal number of rhythm of reference characteristic sequence Ns. And if the two numbers are not equal, entering the step 2.
Step 1402, finding out the maximum public substring Str of the reference characteristic sequence and the scoring characteristic sequence by using a dynamic programming algorithmlcstringAnd its starting position PlcstringAnd at a starting position PlcstringCutting the reference characteristic sequence and the scoring characteristic sequence into a front part and a back part respectively S1first_half、S1second_half、S2first_half、S2second_half。
Step 1403, find out S1 by dynamic programming algorithmfirst_half、S2first_halfMaximum common subsequence of (LCS)first_alfAnd its corresponding position map Pfirst_half。
Step 1404, finding out S1 by dynamic programming algorithmsecond_half、S2second_halfMaximum common subsequence of (LCS)second_halfAnd its corresponding position map Psecond_half。
Step S1405, merging LCSfirst_half、LCSsecond_halfObtaining the whole matched maximum public subsequence LCS, merging Pfirst_half、Psecond_halfObtain the whole matching result Plcs。
A scoring method used in step 15, comprising the steps of:
step 1501, calculating the score of a single rhythm, wherein the total rhythm score weight alpha of the rhythm examination type test questions can be set to be 1, and the number of rhythms of the reference characteristic model is NsThen the score for a single tempo is:
Scores＝100×α÷Ns
step 1502, according to the matching result of the reference characteristic sequence and the scoring characteristic sequence, the number of rhythms matched in the reference characteristic sequence is counted as the hit number NcorrectCalculating the number of unmatched rhythms in the reference characteristic sequence as the number of missed singing NmissCalculating the number of unmatched rhythms in the scoring characteristic sequence as the number of singing NextraIn which N iscorrect+NmissTotal number of rhythm of reference characteristic sequence Ns，NextraTotal number of rhythm of scoring character sequence Nx-total number of reference signature sequences cadences Ns。
Step 1503, calculating a total score:
Figure BDA0002744879080000091
according to a calculation formula, the score of the full score can be obtained only when all rhythms of the reference characteristic sequence are matched and no missed singing or multi-singing rhythms exist, and the score is lower when the number of the matched rhythms is less; the more missed and multiple singing rhythms, the lower the score will be.
Fig. 6 is a flowchart of the extraction and scoring of rhythm examination type audio rhythm and pitch characteristics, which mainly relates to a method for extracting and scoring rhythm examination type audio rhythm and pitch characteristics, and comprises the following steps:
step 20, inputting the title information, the standard rhythm characteristic sequence (rhythm code) and the standard pitch characteristic sequence (pitch code) of the melody examination question, and referring to step S1 in the specific implementation process.
And step 21, recording and uploading the audio file by the singer through a mobile terminal such as a smart phone. The singer checks the test question information, sings the music rhythm towards the mobile phone microphone, and submits and uploads the singing audio. The background server 140 performs denoising preprocessing on the uploaded audio to be scored, and then performs CQT conversion to obtain an audio CQT spectrogram.
And step 22, finding out potential note starting points according to a note starting point detection algorithm, and dividing the spectrogram into a plurality of segments according to the width of a fixed window by taking each note starting point as a center.
And step 23, predicting the segmented spectrogram segments based on the deep learning model, and finding out all the marked note starting points of the whole audio. Since the rhythm examination type theme scoring characteristics comprise two parts of rhythm and pitch, the following steps 24, 25 and 26 finish the scoring of the rhythm characteristics; while steps 27, 28, 29 complete the scoring of the pitch characteristics.
And 24, calculating the duration of each note according to the intervals of all the marked note starting points, and confirming the rhythm type of each note based on the statistical characteristics of the note duration and the continuity of the preceding note and the following note to obtain the rhythm characteristic sequence of the whole audio.
And step 25, matching the rhythm characteristic sequence with a standard rhythm characteristic sequence of the reference test question to find out the number of multi-singing rhythms, the number of missed singing rhythms and a matching alignment result.
And 26, obtaining the score of each note according to the weight of the rhythm and the pitch, calculating a final score according to the matching alignment result, the number of multi-singing rhythms and the number of missed-singing rhythms and a scoring rule, and returning the final score and scoring details.
And 27, finding out the fundamental tone of each note on the audio spectrogram according to all the marked note starting points, and calculating the pitch characteristic sequence of the whole audio according to the fundamental tone of each note.
And step 28, matching the pitch characteristic sequence with the standard pitch characteristic sequence of the reference test question, and finding out the number of multi-singing pitches, the number of missed singing pitches and a matching alignment result.
And step 29, obtaining the score of each note according to the weight of the rhythm and the pitch, calculating a final score according to the matching alignment result, the number of multi-tone pitches, the number of missed tone pitches and a scoring rule, and returning the final score and scoring details.
Fig. 7 is a flowchart of the melody examination type audio note onset segmentation based on deep learning according to the present invention, and the diagram mainly relates to a note onset segmentation method based on deep learning, which includes the following steps:
2301, obtaining a spectrogram of a training sample audio through CQT transformation;
2302, obtaining a note onset vector V for a training sample audio by selecting a peak in the audio amplitude envelope to locate the note onset eventonsetsAnd removing the noise interference event through a threshold value.
2304, passing the note onset vector VonsetsSegmenting a training sample CQT spectrogram into spectrogram segments according to the segmentation rule: by note onset vector VonsetsFrequency sequence Number of each note starting pointfFor the middle of the segmentation, the number of frames occupied by the segment is the Length of the segmentation windoww. And forming a training sample Set through manual markingtrain。
2305, inputting the artificially labeled training samples into a deep learning neural network to judge and learn the note starting point to obtain a training Modeltrain。
2306, obtaining the note starting point vector by the time domain amplitude envelope peak value characteristic
Figure BDA0002744879080000111
Step 2307, obtaining and obtaining the note starting point vector through frequency domain frequency hopping characteristic
Figure BDA0002744879080000112
2308, get the note starting point vector
Figure BDA0002744879080000113
And
Figure BDA0002744879080000114
are combined to obtain
Figure BDA0002744879080000115
And by note onset vectors
Figure BDA0002744879080000116
And segmenting the audio CQT spectrogram to be predicted into spectrogram segments.
Step 2309, loading training ModeltrainInputting CQT spectrogram fragments of the audio to be predicted for prejudging to obtain judgment results of all spectrogram fragments;
23010, obtaining the frame number information corresponding to all the spectrogram segments determined as positive, and obtaining the note starting point frame number vector of the audio to be predicted
Figure BDA0002744879080000117
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.

Claims (8)

1. A music singing scoring system for middle and primary school artistic quality assessment is characterized in that: the system comprises a server, a firewall, a communication network, a computer and a mobile terminal;
the server is a server cluster formed by a plurality of physical devices;
the communication network is a wired network or a wireless network and is mainly used for connecting the server, the computer and the mobile terminal to perform data interaction.
2. The method for scoring a musical performance in a middle and primary school quality of art assessment according to claim 1, comprising the steps of:
step 1: the administrator inputs the standard characteristic sequence of the reference test questions through the computer and uploads and saves the standard characteristic sequence to the server; the reference test questions are mainly divided into two types: rhythm type examination questions and melody type examination questions, wherein the rhythm type examination questions generally only examine the mastering conditions of singers on music rhythm, the standard characteristic sequence and the scoring characteristic sequence of the rhythm type examination questions are music rhythm codes, the melody type examination questions generally simultaneously examine the mastering conditions of singers on music rhythm and music melody, and the standard characteristic sequence and the scoring characteristic sequence of the rhythm type examination questions are music rhythm codes and music pitch codes;
step 2: a singer inputs and uploads an audio file through a mobile terminal such as a smart phone, the singer checks the type of a test question, and if the test question is a rhythm test question, the singer only needs to sing a music rhythm towards a mobile phone microphone; if the test question is melody type test question, the mobile phone microphone needs to be sung music melody, including rhythm and pitch;
and step 3: the server acquires a scoring feature sequence of the audio file to be scored through a feature extraction algorithm, and matches the scoring feature sequence with a standard feature sequence; the rhythm type examination questions only need to be matched with the rhythm characteristic sequence, and the rhythm type examination questions need to be matched with the rhythm characteristic sequence and the pitch characteristic sequence;
and 4, step 4: and 3, calculating a score according to the matching result of the step 3 and a scoring rule, and returning the score and scoring details.
3. The method for scoring a musical performance in a middle and primary school quality of art assessment according to claim 2, wherein: inputting data of reference standard test questions according to the step 1, wherein the data comprises types of the reference standard test questions; inputting rhythm examination type test questions into rhythm duration codes of each bar of standard test questions; the rhythm examination type test question is recorded with rhythm duration code and pitch code of each bar of the standard test question.
4. The method for scoring a musical performance in a middle and primary school quality of art assessment according to claim 1, wherein: and according to the step 2, converting the singing audio of the singer into the wav standard format through the audio format.
5. The method for scoring a musical performance in a middle and primary school quality of art assessment according to claim 4, wherein: and carrying out noise reduction processing on wav standard audio of the singer, and carrying out feature extraction according to the type of the singing test questions. And extracting rhythm characteristics of the rhythm examination type test questions through an audio rhythm characteristic extraction module according to the sound amplitude change condition, and calculating to obtain a rhythm characteristic sequence. The rhythm examination type test question completes note segmentation through the note segmentation prediction analysis module, the duration of each note after segmentation is calculated, the rhythm feature sequence of the whole audio is obtained, and the pitch feature extraction module extracts pitch features of each note after segmentation and obtains the pitch feature sequence of the whole audio.
6. The method for scoring a musical performance in a middle and primary school quality of art assessment according to claim 2, wherein: and matching the obtained rhythm characteristic sequence, pitch characteristic sequence and standard scoring characteristic sequence through a characteristic sequence matching module, and finding out the matching result, the number and the position of the multi-singing notes and the number and the position of the missed-singing notes of each note in the standard scoring characteristic sequence.
7. The method for scoring a musical performance in a middle and primary school quality of art assessment according to claim 2, wherein: the score and result display module calculates the score of each note according to the set score weight, the total score and the standard score characteristic sequence number, further obtains the score of each note according to the matching result of each note, deducts the related score according to the number of the multi-note and the number of the missed note, subtracts the deducted score from the score summation result of all notes to obtain the final score, and simultaneously displays the score details according to the matching result of each note, the number of the multi-note and the number of the missed note.
8. The method for scoring a musical performance in a middle and primary school quality of art assessment according to claim 2, wherein: the method comprises the steps of converting audio into a spectrogram through a CQT algorithm, dividing original data samples according to a time sequence by taking the wave trough of an envelope line as a center and a fixed window width in combination with a music amplitude envelope line wave trough detection algorithm to form small segments of the spectrogram, and marking positive and negative samples in a mode of manually judging whether the small segments of the spectrogram contain note starting points. And training positive and negative samples by adopting deep learning, and storing the trained network model.
CN202011162757.2A 2020-10-27 2020-10-27 Music singing scoring system and method for evaluating artistic quality of primary and secondary schools Pending CN112201100A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011162757.2A CN112201100A (en) 2020-10-27 2020-10-27 Music singing scoring system and method for evaluating artistic quality of primary and secondary schools

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011162757.2A CN112201100A (en) 2020-10-27 2020-10-27 Music singing scoring system and method for evaluating artistic quality of primary and secondary schools

Publications (1)

Publication Number Publication Date
CN112201100A true CN112201100A (en) 2021-01-08

Family

ID=74011508

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011162757.2A Pending CN112201100A (en) 2020-10-27 2020-10-27 Music singing scoring system and method for evaluating artistic quality of primary and secondary schools

Country Status (1)

Country Link
CN (1) CN112201100A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114783222A (en) * 2022-05-11 2022-07-22 张�浩 Musical instrument auxiliary teaching method and system based on AR technology

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102664016A (en) * 2012-04-23 2012-09-12 安徽科大讯飞信息科技股份有限公司 Singing evaluation method and system
TW201239876A (en) * 2011-03-17 2012-10-01 Univ Nat Taipei Technology Scoring system for singing
CN103971674A (en) * 2014-05-22 2014-08-06 天格科技（杭州）有限公司 Sing real-time scoring method accurate in scoring and good in user experience
CN104064180A (en) * 2014-06-06 2014-09-24 百度在线网络技术（北京）有限公司 Singing scoring method and device
CN107103912A (en) * 2017-04-24 2017-08-29 行知技术有限公司 A kind of student for imparting knowledge to students and checking and rating sings performance points-scoring system
CN107785010A (en) * 2017-09-15 2018-03-09 广州酷狗计算机科技有限公司 Singing songses evaluation method, equipment, evaluation system and readable storage medium storing program for executing
CN107818796A (en) * 2017-11-16 2018-03-20 重庆师范大学 A kind of music exam assessment method and system
CN109448754A (en) * 2018-09-07 2019-03-08 南京光辉互动网络科技股份有限公司 A kind of various dimensions singing marking system
CN111369975A (en) * 2020-03-17 2020-07-03 郑州工程技术学院 University music scoring method, device, equipment and storage medium based on artificial intelligence

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TW201239876A (en) * 2011-03-17 2012-10-01 Univ Nat Taipei Technology Scoring system for singing
CN102664016A (en) * 2012-04-23 2012-09-12 安徽科大讯飞信息科技股份有限公司 Singing evaluation method and system
CN103971674A (en) * 2014-05-22 2014-08-06 天格科技（杭州）有限公司 Sing real-time scoring method accurate in scoring and good in user experience
CN104064180A (en) * 2014-06-06 2014-09-24 百度在线网络技术（北京）有限公司 Singing scoring method and device
CN107103912A (en) * 2017-04-24 2017-08-29 行知技术有限公司 A kind of student for imparting knowledge to students and checking and rating sings performance points-scoring system
CN107785010A (en) * 2017-09-15 2018-03-09 广州酷狗计算机科技有限公司 Singing songses evaluation method, equipment, evaluation system and readable storage medium storing program for executing
CN107818796A (en) * 2017-11-16 2018-03-20 重庆师范大学 A kind of music exam assessment method and system
CN109448754A (en) * 2018-09-07 2019-03-08 南京光辉互动网络科技股份有限公司 A kind of various dimensions singing marking system
CN111369975A (en) * 2020-03-17 2020-07-03 郑州工程技术学院 University music scoring method, device, equipment and storage medium based on artificial intelligence

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
邓志勇等: "《音乐声学导论》", 28 February 2019, 首都师范大学出版社 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114783222A (en) * 2022-05-11 2022-07-22 张�浩 Musical instrument auxiliary teaching method and system based on AR technology
CN114783222B (en) * 2022-05-11 2022-12-20 张�浩 Musical instrument auxiliary teaching method and system based on AR technology

Similar Documents

Publication Publication Date Title
CN102664016B (en) Singing evaluation method and system
CN106485984B (en) Intelligent teaching method and device for piano
CN103824565A (en) Humming music reading method and system based on music note and duration modeling
CN111081272A (en) Song climax fragment identification method and device
Lee et al. A Unified System for Chord Transcription and Key Extraction Using Hidden Markov Models.
CN110364184B (en) Intonation evaluation method based on deep convolutional neural network DCNN and CTC algorithm
CN116206496B (en) Oral english practice analysis compares system based on artificial intelligence
Bozkurt et al. A dataset and baseline system for singing voice assessment
CN106935236A (en) A kind of piano performance appraisal procedure and system
CN107767850A (en) A kind of singing marking method and system
CN105895079B (en) Voice data processing method and device
Zhang et al. Automatic singing evaluation without reference melody using bi-dense neural network
Köktürk-Güzel et al. Automatic assessment of student rhythmic pattern imitation performances
CN112201100A (en) Music singing scoring system and method for evaluating artistic quality of primary and secondary schools
Pikrakis et al. Tracking melodic patterns in flamenco singing by analyzing polyphonic music recordings
Schramm et al. Automatic Solfège Assessment.
CN112634841A (en) Guitar music automatic generation method based on voice recognition
CN111785236A (en) Automatic composition method based on motivational extraction model and neural network
Falcao et al. A dataset of rhythmic pattern reproductions and baseline automatic assessment system
CN114678039A (en) Singing evaluation method based on deep learning
KR20190112885A (en) Automated estimation method for music dictation and estimation system using the same
CN111210845B (en) Pathological voice detection device based on improved autocorrelation characteristics
Chen Characterization of pitch intonation of Beijing opera
Tang et al. Melody Extraction from Polyphonic Audio of Western Opera: A Method based on Detection of the Singer's Formant.
Shirali-Shahreza et al. Fast and scalable system for automatic artist identification

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication

Application publication date: 20210108

RJ01 Rejection of invention patent application after publication