timbre_range / README.md
admin
test cls label
242afee
|
raw
history blame
5.14 kB
metadata
license: mit
task_categories:
  - audio-classification
language:
  - zh
  - en
tags:
  - music
  - art
pretty_name: Timbre and Range Dataset
size_categories:
  - 1K<n<10K
dataset_info:
  - config_name: timbre
    features:
      - name: audio
        dtype:
          audio:
            sampling_rate: 44100
      - name: mel
        dtype: image
      - name: label
        dtype:
          class_label:
            names:
              '0': Base
              '1': Split
              '2': Short
      - name: score1
        dtype: float64
      - name: score2
        dtype: float64
      - name: avg_score
        dtype: float64
    splits:
      - name: train
        num_bytes: 213644
        num_examples: 537
      - name: validation
        num_bytes: 26664
        num_examples: 67
      - name: test
        num_bytes: 27088
        num_examples: 68
    download_size: 595425921
    dataset_size: 267396
  - config_name: range
    features:
      - name: audio
        dtype:
          audio:
            sampling_rate: 44100
      - name: mel
        dtype: image
      - name: label
        dtype:
          class_label:
            names:
              '0': Narrow
              '1': Moderate
              '2': Wide
    splits:
      - name: train
        num_bytes: 210052
        num_examples: 580
      - name: validation
        num_bytes: 26462
        num_examples: 73
      - name: test
        num_bytes: 26400
        num_examples: 73
    download_size: 65309164
    dataset_size: 262914
configs:
  - config_name: timbre
    data_files:
      - split: train
        path: timbre/train/data-*.arrow
      - split: validation
        path: timbre/validation/data-*.arrow
      - split: test
        path: timbre/test/data-*.arrow
  - config_name: range
    data_files:
      - split: train
        path: range/train/data-*.arrow
      - split: validation
        path: range/validation/data-*.arrow
      - split: test
        path: range/test/data-*.arrow

Dataset Card for Timbre and Range Dataset

Dataset Summary

The timbre dataset contains acapella singing audio of 9 singers, as well as cut single-note audio, totaling 775 clips (.wav format) The vocal range dataset includes several up and down chromatic scales audio clips of several vocals, as well as the cut single-note audio clips (.wav format).

Supported Tasks and Leaderboards

Audio classification

Languages

Chinese, English

Dataset Structure

https://huggingface.co/datasets/ccmusic-database/timbre_range/viewer

Data Instances

.zip(.wav, .jpg), .csv

Data Fields

timbre: song1-32
range: vox1_19-22/26-29/32/33/36-38/41-47/51-55/59-64/69-71/79-81

Data Splits

Train, Validation, Test

Usage

Timbre subset

from datasets import load_dataset

ds = load_dataset("ccmusic-database/timbre_range", name="timbre")
for item in ds["train"]:
    print(item)

for item in ds["validation"]:
    print(item)

for item in ds["test"]:
    print(item)

Range subset

from datasets import load_dataset

ds = load_dataset("ccmusic-database/timbre_range", name="range")
for item in ds["train"]:
    print(item)

for item in ds["validation"]:
    print(item)

for item in ds["test"]:
    print(item)

Maintenance

GIT_LFS_SKIP_SMUDGE=1 git clone [email protected]:datasets/ccmusic-database/timbre_range
cd timbre_range

Mirror

https://www.modelscope.cn/datasets/ccmusic-database/timbre_range

Dataset Creation

Curation Rationale

Promoting the development of music AI industry

Source Data

Initial Data Collection and Normalization

Zijin Li, Zhaorui Liu, Monan Zhou

Who are the source language producers?

Composers of the songs in dataset

Annotations

Annotation process

CCMUSIC students collected acapella singing audios of 9 singers, as well as cut single-note audio, totaling 775 clips

Who are the annotators?

Students from CCMUSIC

Personal and Sensitive Information

Due to copyright issues with the original music, only acapella singing audios are provided in the dataset

Considerations for Using the Data

Social Impact of Dataset

Promoting the development of AI in the music industry

Discussion of Biases

Most are Chinese songs

Other Known Limitations

Samples are not balanced enough

Additional Information

Dataset Curators

Zijin Li

Evaluation

[1] Yiliang, J. et al. (2019) 'Data Augmentation based Convolutional Neural Network for Auscultation', Journal of Fudan University(Natural Science), pp. 328-334. doi:10.15943/j.cnki.fdxb-jns.2019.03.004.

Citation Information

@article{2019Data,
  title={Data Augmentation based Convolutional Neural Network for Auscultation},
  author={Yiliang Jiang and Xulong Zhang and Jin Deng and Wenqiang Zhang and Wei Li},
  journal={Journal of Fudan University (Natural Science)},
  year={2019},
}

Contributions

Provide a dataset for music timbre and range