China Sound and Music Technology Database collects pop music, folk music and the sound materials of national musical instruments, and makes comprehensive annotation to form a multi-purpose music database for MIR researchers. This database is specially recorded by the conservatory of music, the recordists have high music literacy, professional recording environment and technology, high recording quality, no commodity copyright problems, and the recorded audio is free and public, which is convenient for large-scale promotion. This database professionally limits the recording environment, recording equipment, recording personnel and processes, so as to avoid variety noises interference and obtain high-quality audio materials. In addition, it is of great significance to the research of music information retrieval to record the melody part and accompaniment part independently. In the future, we will collect more music materials for recording and detailed annotation.
This database contains 5 sub-databases, which are listed as follows:
Based on the working idea of combining manual labeling with computer in the construction of World Music Database, this database collects and labels the audio of five modes (including five tones, six tones and seven tones) of "Gong, Shang, Jue, Zhi and Yu". At the same time, it makes a detailed analysis of the judgment of Chinese national pentatonic modes, and finds application scenarios and technical models, which can provide raw data for the analysis and retrieval of Chinese national music characteristics.
Some audio lists are as follows:
| Mode Type |
Name | Performer | Album Name |
National Mode Name |
Tonggong System |
Audio Links |
|---|---|---|---|---|---|---|
| Gong mode | Liang Xiao Yin | Gong Yi | Gu Qin Yan Zou Fa | F Gong/Pentatonic mode | F Gong System | Liang Xiao Yin |
| Shang mode | Jin She Kuang Wu | Liu De Hai | Pipa | E Shang/Seven tone mode/Qing Yue | D Gong System | Jin She Kuang Wu |
| Jue mode | Hu Jia Shi Ba Po | China National Traditional Orchestra | China Chun Guzheng | D Jue/Seven tone mode/Qing Yue | B Flat Gong System | Hu Jia Shi Ba Po |
| Zhi mode | Zui Yu Chang Wan | Gong Yi | Gong Yi Gu Qin | G Zhi/Seven tone mode/Qing Yue | C Gong System | Zui Yu Chang Wan |
| Yu mode | Ping Sha Luo Yan | Guan Ping Hu | Gu Qin | E Flat Yu/Pentatonic mode | G Flat Gong System | Ping Sha Luo Yan |
This database contains about 300 Guzheng audios, which can be used for data-driven automatic music transcription and translation related algorithms. The production process is to first collect all kinds of music scores, make MusicXML through music arrangement software, and then convert it to MIDI. On this basis, for the technical representation, we debug the sound in MIDI through the plug-in that changes the tone in real time to get the glide and vibrato. The scraping remote refers to the use of irregular continuous scales (8 ~ 12 continuous tones) to compose and express in MIDI, and the arpeggio directly plays four associated tones at the same time through the chord relationship.
The total length of all audios is about 20 hours, with about 150000 annotation, including onset, offset, speed and notes. All Guzheng tracks are in wav and MIDI format, which provides some limited help for the transcription and translation of traditional musical instruments.
Some data lists are as follows:
This database contains 2824 audio clips of guzheng playing techniques. Among them, 2328 pieces were collected from virtual sound banks, and 496 pieces were played and recorded by a professional guzheng performer. These clips cover almost all the tones in the range of guzheng and the most commonly used playing techniques in guzheng performance. According to the different playing techniques of guzheng, the clips are divided into 8 categories: Vibrato(chanyin), Upward Portamento(shanghuayin), Downward Portamento(xiahuayin), Returning Portamento(huihuayin), Glissando (guazou, huazhi), Tremolo(yaozhi), Harmonic(fanyin), Plucks(gou,da,mo,tuo…).
Some data lists are as follows:
Music Emotion Recognition (MER) has recently received considerable attention. To support the MER research which requires large music content libraries, we present the PMEmo dataset containing emotion annotations of 794 songs as well as the simultaneous electrodermal activity (EDA) signals. A Music Emotion Experiment was well-designed for collecting the affective-annotated music corpus of high quality, which recruited 457 subjects. The dataset is publically available to the research community, which is foremost intended for benchmarking in music emotion retrieval and recognition. To straightforwardly evaluate the methodologies for music affective analysis, it also involves pre-computed audio feature sets. In addition to that, manually selected chorus excerpts (compressed in MP3) of songs are provided to facilitate the development of chorus-related research. In Our article, The PMEmo Dataset for Music Emotion Recognition, We describe in detail the resource acquisition, subject selection, experiment design and annotation collection procedures, as well as the dataset content and data reliability analysis. We also illustrate its usage in some simple music emotion recognition tasks which testified the PMEmo dataset’s competence for the MER work. Compared to other homogeneous datasets, PMEmo is novel in the organization and management of the recruited annotators, and it is also characterized by its large amount of music with simultaneous physiological signals.
Some data in the dataset are listed as follows:
| Audio Serial | Song Metadata | Audio Demo | Pre-computed Audio Features for Use in MER Tasks | Manually Annotated Emotion Labels | EDA Physiological Signals | Song Lyrics (LRC) | Song Comments |
|---|---|---|---|---|---|---|---|
| 1 | |||||||
| 5 | |||||||
| 6 | |||||||
| 9 |
Instrument playing technique (IPT) is a key element of musical presentation.
Guzheng is a polyphonic instrument. In Guzheng performance, notes with different IPTs are usually overlapped and mixed IPTs that can be decomposed into multiple independent IPTs are usually used. Most existing work on IPT detection typically uses datasets with monophonic instrumental solo pieces. This dataset fills a gap in the research field.
The dataset comprises 99 Guzheng solo compositions, recorded by professionals in a studio, totaling 9064.6 seconds. It includes seven playing techniques labeled for each note (onset, offset, pitch, vibrato, point note, upward portamento, downward portamento, plucks, glissando, and tremolo), resulting in 63,352 annotated labels. The dataset is divided into 79, 10, and 10 songs for the training, validation, and test sets, respectively.
More details about the code and datasets can be found at https://lidcc.github.io/GuzhengTech99/
The paper will be released at ICASSP 2023 (IEEE International Conference on Acoustics, Speech, and Signal Processing).
Label samples：
| onset_time | offset_time | IPT | note |
|---|---|---|---|
| 2.252335601 | 3.018594104 | boxian | 69 |
| 3.018594104 | 3.81968254 | boxian | 57 |
| 3.81968254 | 4.551111111 | boxian | 59 |
| 4.551111111 | 5.317369615 | boxian | 47 |
| 5.317369615 | 6.025578231 | chanyin | 62 |
| 6.025578231 | 6.397097506 | boxian | 50 |
| 6.397097506 | 6.745396825 | chanyin | 62 |
| 6.745396825 | 7.476825397 | boxian | 50 |
| 7.476825397 | 8.196643991 | chanyin | 62 |
| 8.196643991 | 8.93968254 | shanghua | 59 |