TW201426729A - Automatic emotion classification system with gamut-type sound effects - Google Patents

Abstract

The present invention provides an automatic emotion classification system with gamut-type sound effects, which may perform emotion classification for gamut-type sound effects stored in an electronic device. The automatic emotion classification system comprises a sound effect classification library, a sound effect reading module, and a tone analysis module, a tone color analysis module and a sound effect emotion classification module for respectively analyzing the gamut-type sound effects. A user may first establish a design of multi-type emotion prediction model specifically for the user based on his or her listening feeling, so that the emotion classification for gamut-type sound effects may be more approached to the user's preference. Thus, the present invention may satisfy all kinds of user's requirements, and present a novel gamut-type sound effect classification design.

Description

Automatic emotion classification system with scale-like sound effects

The present invention relates to a sound classification system, and more particularly to an emotion classification system with scale-like sound effects.

With the popularity of mobile devices such as smart phones and tablets, and the diversity of communication software, modern people use these mobile devices and communication software to communicate and share interesting things with family members or friends almost every day. For example, in the transmitted image newsletter, the sound effect is included, or the film is uploaded to the Internet for sharing, but it is often found that when the sound is to be carried in the image or the film, most of the built-in system sound effects can be used only, and the selection is less. And more tedious. Therefore, users of mobile devices usually take music-like sound effects such as music clips or video game sound effects as background music for their short messages or short films, or as ringing for incoming calls, and even as a background sound for the developed video game. But which kind of sound effect should be used to generate an emotional response to the newsletter, film or game content, or which sound effect is not happy, sad or passionate, so it usually takes a lot of time to try. More time consuming. Although many companies are currently developing software systems that can classify emotions in music or sound effects, the current system is a systematic classification method that does not consider everyone's emotions for sound effects. Feel the difference and poor applicability.

Accordingly, it is an object of the present invention to provide an automatic emotion classification system that can classify emotions with scales based on user preferences.

Therefore, the automatic emotion classification system with scale-like sound effects of the present invention can be used to classify emotions with scale-like sound effects, and the automatic emotion classification system includes a sound effect classification library, an audio effect reading module, and a tonal analysis module. A tone color analysis module and a sound effect emotion classification module. The sound effect classification library has a plurality of emotional sound effect units respectively configured with an emotional adjective parameter representing a specific emotional adjective, and the emotional sound effect units are respectively configured to store the sounded sound effects. The sound reading module can be driven to read each of the scale-like sound effects stored by the emotional sound unit. The tonality analysis module can analyze all the tonalities contained in each of the scale-like sound effects read by the sound effect reading module, and statistically analyze all the tonalities contained in each of the scale-like sound effects, and classify them into large The proportion of each tonality of tonality, the proportion of each tonality classified as minor tonality, the total proportion of all tonality covered by majority, and all tonality covered by minority The total score is proportional to the output one-tone parameter. The timbre analysis module can analyze all the timbre-like sounds read by the sound-effect reading module, and output a timbre parameter corresponding to each of the scale-like sound effects. The sound emotion classification module includes a classification model establishing unit and a classification storage unit, and the classification model establishing unit analyzes and analyzes all the tonal parameters and the tone parameters of the scale-like sound effects stored in each emotional sound unit, and cooperates with the Emotional sound unit Emotional adjective parameters, and a multi-category emotion prediction model is established, and the multi-category emotion classification model can be executed to analyze the newly acquired scale-like sound effects of the electronic device and associate it with one of the emotional adjective parameters. Pairing, the classified storage unit stores the newly captured scaled sound effects into the emotional sound effect unit corresponding to the paired emotional adjective parameters.

The effect of the invention: the user can first establish a design of a multi-category emotion prediction model based on his listening experience, so that the emotion classification with the scale-like sound effect will be closer to the user preference, so that the invention Being able to meet the needs of all types of users is an innovative sound classification design.

3‧‧‧Audio reading module

5‧‧‧Sound classification library

51‧‧‧Emotional sound unit

6‧‧‧Tense Analysis Module

7‧‧‧ tone color analysis module

8‧‧‧Sound Emotion Classification Module

81‧‧‧Classification model building unit

82‧‧‧Classified storage unit

900‧‧‧Electronic devices

902‧‧‧ Speaker

Other features and effects of the present invention will be apparent from the following description of the drawings, wherein: FIG. 1 is a functional block diagram of a preferred embodiment of the automatic emotion classification system with scale-like sound effects of the present invention.

As shown in FIG. 1 , the automatic emotion classification system with scale-like sound effects of the present invention can be used to classify emotions with scales and can be used in an electronic device 900. The electronic device 900 can be a smart phone. A mobile device such as a tablet or a notebook computer may be a desktop computer or the like, and is not limited to the above types. The sound effects of the scales generally refer to sound effects such as various musical performances, electronic music, video game music, and the like.

The automatic emotion classification system comprises an audio effect reading module 3, a sound effect classification library 5, a tone analysis module 6, a tone color analysis module 7, and a sound effect emotion classification module 8.

The sound effect reading module 3 can be driven to read the sound effects of the scales stored in the sound effect classification library 5, and transmit all the sound effects of the scales to the tonality analysis module 6 and the sound color analysis module. 7 The analysis is performed. In addition, the sound effect reading module 3 can also be driven to read the sound effect of the scaled sound, and the sound output is output through a speaker 902 of the electronic device 900 for listening.

The sound effect classification library 5 has a plurality of emotional sound effect units 51, and the emotional sound effect units 51 respectively set an emotional adjective parameter corresponding to a certain kind of emotional adjective, and can respectively be used by the user of the electronic device 900 to listen to the emotional feelings of the user. The currently stored audio-like sound effects of the electronic device 900 are respectively stored in the emotional sound effect unit 51 corresponding to an emotional adjective for preliminary classification and storage.

The tonality analysis module 6 can analyze all the sound effects of the scales read by the sound effect reading module 3, and perform tonal analysis on each sound effect of the scales, and analyze all the tones contained in the sound effects of each scale. Sexuality, and statistical analysis of the proportion of each tonality classified as major tonality, the proportion of each tonality classified as minor tonality, and the total proportion of all tonality covered by majority. And the total score of all tonality covered by the minor tonality, and the corresponding output tone parameter.

The timbre analysis module 7 can analyze all the timbre-like sounds read by the sound effect reading module 3, and output a timbre data corresponding to each of the gradation-like sound effects.

In the implementation, because the tonality and timbre analysis of the scale-like sound effects are prior art, and the analysis methods and related devices are numerous, for example, by micro-processing wafers with MATLAB programs or other related analysis programs, etc., they are not detailed. .

The sound effect emotion classification module 8 includes a classification model establishing unit 81 and a classification storage unit 82. The classification model establishing unit 81 statistically analyzes the tonal parameters and the timbre parameters of all the scale-like sound effects stored in each of the emotional sound effects units 51 through a multi-category support vector machine (SVM), and according to the emotional sound effect units. 51 sets the emotional adjective parameters, and establishes a multi-category emotion prediction model. The multi-category emotion prediction model can be executed to perform sentiment classification analysis on the scaled sound effects newly captured or stored by the electronic device 900, and classify the newly captured scale-like sound effects into one of the emotions. Adjective parameters. The classification storage unit 82 can store the new captured scale-like sound effects in the emotional sound effect unit 51 corresponding to the emotional adjective parameters.

Through the above design, after the electronic device 900 is installed and installed with the automatic emotion classification system of the present invention, the user can first classify and store the existing sound-like sound effects in the electronic device 900 in a corresponding emotional adjective according to the listening experience. In the emotional sound effect unit 51, the automatic emotion classification system is activated. At this time, the automatic emotion classification system first performs tonal analysis and timbre analysis according to all the scale-like sound effects currently stored in the emotional sound effect unit 51. And establish a proprietary multi-category emotion prediction model.

After the multi-category emotion prediction model is established, when the electricity is After the sub-device 900 captures or stores a new sound effect with a scale, the sound effect reading module 3 can read the new sound level sound effect, and the tonality analysis module 6 and the sound color analysis module 7 The tonal parameters and the timbre parameters are analyzed, and the multi-category emotion prediction model is used to analyze the emotional adjective parameters corresponding to the new gradation-like sound effects according to the tonal parameters and the timbre parameters. At this time, the classification storage unit 82 automatically stores the new scale-like sound effect into the corresponding emotion sound effect unit 51, and automatically completes the scale-like sound effect emotion classification.

In the future, when users need to use the sound with scale, for example, to bring a sound with a scale in the newsletter, or to bring a sound with a scale in the produced movie or game, you can first according to the scale you want to entrain. The sound effect of the sound effect, the sound effect reading module 3 directly reads the sound effects of the scales that have been stored in the corresponding emotional sound effect unit 51 through the automatic emotion classification system of the present invention, and outputs the listening sound through the speaker 902 to select The desired sound with a scale is quite convenient and practical.

In summary, the automatic emotion classification system classification of the present invention allows the user to first classify the existing scale-like sound effects of the electronic device 900 into the emotional sound effect units 51 according to the listening experience, and then analyze the sound effects successively. The tonality parameter and the timbre parameter are used to establish a design of a multi-category emotion prediction model that is exclusive to oneself. When the electronic device 900 newly captures or stores a new band-like sound effect, the new band can be analyzed. The tonal parameters of the scale-like sound effects and the timbre parameters, and the multi-category emotion prediction model directly analyzes the emotional-adjective parameters of the new-scale sound effects, since the multi-category emotion prediction model is based on the user's sound effects with the scales The listening experience is established, so the emotional classification result with the scale sound effect will be closer to the user's preference, which is quite convenient. Therefore, with the above design, the automatic emotion classification system with the sound level of the present invention can be used by the user to establish a unique sound emotion classification rule according to his preference, and can satisfy the needs of various types of users, which is an innovation. The classification of sound effects with scales. Therefore, the object of the present invention can be achieved.

The above is only the preferred embodiment of the present invention, and the scope of the present invention is not limited thereto, that is, the simple equivalent changes and modifications made by the patent application scope and patent specification content of the present invention, All remain within the scope of the invention patent.

3‧‧‧Audio reading module

5‧‧‧Sound classification library

51‧‧‧Emotional sound unit

6‧‧‧Tense Analysis Module

7‧‧‧ tone color analysis module

8‧‧‧Sound Emotion Classification Module

81‧‧‧Classification model building unit

82‧‧‧Classified storage unit

900‧‧‧Electronic devices

902‧‧‧ Speaker

Claims (2)

An automatic emotion classification system with scale-like sound effects, which can be used for emotional classification of sound effects with scales, comprising: a sound effect classification library having a plurality of emotional sound effect units respectively configured with an emotional adjective parameter representing a specific emotional adjective, and The emotional sound effect unit is respectively configured to store the sound effects with the scales; an audio effect reading module can be driven to read the sound effects of each of the scales stored by the emotional sound effects units; and a tonal analysis module can analyze the sound effects The sound effect reading module reads all the tonality contained in each of the scale-like sound effects, and statistically analyzes all the tonality contained in each of the scale-like sound effects, and the division of each tonality classified into the major tonality Proportion, the proportion of each tonality classified as minor tonality, the total proportion of all tonality covered by major tonality, and the total proportion of all tonality covered by minor tonality, and the corresponding output Tonality parameter; a tone color analysis module that analyzes all the sounds of the scale-like sounds read by the sound effect reading module, and outputs a tone color parameter corresponding to each band-like sound effect And a sound effect emotion classification module, comprising a classification model establishing unit, and a classification storage unit, the classification model establishing unit will analyze and analyze all the tonal parameters and the tone parameters of the scale-like sound effects stored in each emotional sound unit. And a multi-category emotion prediction model is established in conjunction with the emotional adjective parameters of the emotional sound unit, and the multi-category emotion prediction model can be executed to take a new sound for the electronic device. The class sound effect is analyzed and attributed to one of the emotional adjective parameters, and the classification storage unit stores the newly captured band-like sound effect into the emotional sound effect unit corresponding to the emotional adjective parameter. An automatic emotion classification system with a scale-like sound effect as claimed in claim 1, wherein the classification model establishing unit analyzes, by the multi-category support vector machine, all tonal parameters and sounds of the scale-like sounds stored in each emotional sound unit. The multi-category emotion prediction model is established by parameters and with the emotional adjective parameters of the emotional sound units.