CTMusic: A Traditional Chinese Instrumental Music Dataset Towards Text-to-Music Generation
Contents
Dataset Description
CTMusic is a Chinese traditional instrumental music dataset specifically designed for text-to-music generation. It comprises 741 meticulously annotated text-music pairs with a total duration of 42+ hours, covering a diverse range of traditional Chinese instruments in both solo and ensemble settings. The solo subset covers 12 traditional Chinese instruments, namely bamboo flute (dizi), drum (gu), erhu, guan, guqin, guzheng, pipa, sheng, suona, xiao, xun, and yangqin.
| Subset | #Samples | Hours | Mean Duration (s) | Mean Description Length (words) |
|---|---|---|---|---|
| Solo Set | 601 | 35.59 | 213.19 | 25 |
| Ensemble Set | 140 | 7.11 | 182.76 | 115 |
| Total | 741 | 42.70 | 207.44 | 42 |
Comparison with Other Chinese Music Datasets
| Dataset | Main Task | Format | Annotation | Size |
|---|---|---|---|---|
| ChMusic | Instrument Recognition | Audio | label | 1.4 hours |
| CCMusic | Instrument recognition, Playing technique classification Mode classification, Singing style classification |
Audio | label | 56 hours |
| FolkDB | Symbolic Music Generation | MIDI | - | 11 hours |
| MGD | Song Generation | MIDI | - | - |
| CBFdataset | Playing technique detection | Audio | label | 2.6 hours |
| CCOM-HuQin | Playing technique detection | Audio, Video, Score | label | 4.3 hours |
| JinYue Database | Emotion, scene, and imagery recognition | Audio | label | - |
| GQ39 | Playing technique detection, Mode detection | Audio | label | 0.6 hours |
| CTMusic | Text-to-Music Generation | Audio | text description | 42.7 hours |
Note: The "label" in the table varies depending on the task, such as instrument types, playing techniques, and emotions. CCMusic integrates six existing datasets, including CTIS, GZ IsoTech, Guzheng Tech99, Erhu Playing Technique, CNPM and Bel Canto & Chinese Folk Singing. A considerable portion of its content consists of vocal singing or short audio clips.
Annotated Audio Samples
Annotation: "This traditional Chinese composition showcases the high-pitched suona's vibrant melody, backed by the sheng's steady harmonic chords. The suona plays bright, sustained notes with playful descending slides, following repetitive, symmetrical phrases, while the sheng maintains a consistent rhythmic pulse; the tempo quickens slightly with denser runs, then decelerates as the suona holds a prolonged note, fading with the sheng into silence. Its dynamic shifts and vivid timbral contrasts reflect both folk vitality and traditional structural discipline."
Annotation: "This excerpt begins with a percussion ensemble, followed by a melodic section led by sheng and bamboo flutes. The tempo gradually accelerates, anchored by sustained wind instrument notes. The percussion introduces the piece through syncopated drum and cymbal patterns. The sheng and bamboo flutes then enter with a unison melody dominated by elongated and resonant tones. As the music progresses, the tempo steadily increases. Percussion shifts to denser rhythmic layers, reinforcing the momentum. The final section abruptly halts during a full ensemble unison. The music merges meditative melodic continuity with rhythmic urgency, reflecting traditional music's ceremonial essence."
Annotation: "This is a piece music with a cheerful style. The music is played in a combination of Chinese gongs and drums, bamboo flute, guan and other traditional Chinese instruments. It continues the regularity and melodic quality of ancient music and integrates the sounds of multiple instruments skillfully. The guan plays the main melody, and the gongs and drums run through the whole song as accompaniment. Under the cheerful style, it integrates the liveliness and energy of folk flavor, making the audience feel as if they are in a grand festival."
Annotation: "The music piece is played with sheng, bamboo flutes, drums, cymbals, guan, and yinluo. It starts with a forceful synchronized strike from the drum and yinluo. The sheng and guan follow with a unison melody composed of sustained, resonant notes, supported by sparse cymbal accents. Midway, the bamboo flute emerges as the focal point, performing extended melodic phrases while the sheng and guan shift to harmonic accompaniment. The percussion maintains a consistent pulse with occasional drum rolls and cymbal swells. The final section reunites all instruments in a synchronized restatement of the main melody. The music combines ceremonial solemnity with restrained rhythmic energy, making it suitable for traditional rituals or formal cultural ceremonies."
Annotation: "The piece features bamboo flute, sheng, drums, gongs, and zheng. It starts with a spirited unison opening and transitions into flute-led melodies supported by sheng harmonies. Drums maintain a steady pulse with occasional gong and cymbal accents, while the zheng's tremolos sustain each phrase. Phrases repeat symmetrically with minimal variation. In the finale, the tempo eases and the ensemble restates the melody together, ending on a unified percussive strike. The music exudes a festive and jubilant mood."
Annotation: "This music piece features a solo performance by the Chinese bamboo flute. The melody is highly ornate, decorated with frequent grace notes and rapid trills that showcase the flute's versatility. The rhythmic progression is dynamic; while starting with a moderate flow, the tempo occasionally undergoes sudden, sharp accelerations. It evokes a sense of exuberance, reminiscent of birds chirping in a spring forest or a festive celebration."
Annotation: "This recording features the suona as the lead instrument. The piece maintains a moderate and steady tempo throughout, providing a stable rhythmic foundation. The timbre is notably thin and sharp, cutting through the acoustic space with high-frequency energy. The melody progresses linearly. It is suitable for a scene describing rural life or a folk gathering in a village."
Annotation: "This piece captures the distinctively breathy and hollow sound of the xun. The tempo is exceedingly slow, characterized by long, sustained notes that emphasize the instrument's earthy resonance and audible breath sounds. The pitch range is very narrow, creating a drone-like or chanting texture with almost no melodic modulation or pitch shifting. It evokes a feeling of mystery, desolation, or historical reflection."
Dataset Access and License
The CTMusic dataset, including text descriptions and audio URLs, is publicly available here.
Following established practices in audio research (e.g., AudioSet and MusicCaps), we release URLs rather than audio files to respect the copyright of original recordings.
The dataset is licensed under CC BY-NC 4.0 for research purposes.
Representative Generated Audio Samples
| Text Prompt | SA-pretrained | SA-LoRA | SA-TTT |
|---|---|---|---|
| This traditional Chinese composition showcases the high-pitched suona's vibrant melody, backed by the sheng's steady harmonic chords. The suona plays bright, sustained notes with playful descending slides, following repetitive, symmetrical phrases, while the sheng maintains a consistent rhythmic pulse; the tempo quickens slightly with denser runs, then decelerates as the suona holds a prolonged note, fading with the sheng into silence. Its dynamic shifts and vivid timbral contrasts reflect both folk vitality and traditional structural discipline. | |||
| This is a piece music with a cheerful style. The music is played in a combination of Chinese gongs and drums, bamboo flute, guan and other traditional Chinese instruments. It continues the regularity and melodic quality of ancient music and integrates the sounds of multiple instruments skillfully. The guan plays the main melody, and the gongs and drums run through the whole song as accompaniment. Under the cheerful style, it integrates the liveliness and energy of folk flavor, making the audience feel as if they are in a grand festival. | |||
| The music piece is played with sheng, bamboo flutes, drums, cymbals, guan, and yinluo. It starts with a forceful synchronized strike from the drum and yinluo. The sheng and guan follow with a unison melody composed of sustained, resonant notes, supported by sparse cymbal accents. Midway, the bamboo flute emerges as the focal point, performing extended melodic phrases while the sheng and guan shift to harmonic accompaniment. The percussion maintains a consistent pulse with occasional drum rolls and cymbal swells. The final section reunites all instruments in a synchronized restatement of the main melody. The music combines ceremonial solemnity with restrained rhythmic energy, making it suitable for traditional rituals or formal cultural ceremonies. | |||
| This recording features the suona as the lead instrument. The piece maintains a moderate and steady tempo throughout, providing a stable rhythmic foundation. The timbre is notably thin and sharp, cutting through the acoustic space with high-frequency energy. The melody progresses linearly. It is suitable for a scene describing rural life or a folk gathering in a village. | |||
| This piece captures the distinctively breathy and hollow sound of the xun. The tempo is exceedingly slow, characterized by long, sustained notes that emphasize the instrument's earthy resonance and audible breath sounds. The pitch range is very narrow, creating a drone-like or chanting texture with almost no melodic modulation or pitch shifting. It evokes a feeling of mystery, desolation, or historical reflection. |