Carnatic Varnam Dataset

Carnatic varnam dataset is a collection of 28 solo vocal recordings, recorded for our research on intonation analysis of Carnatic raagas. The collection has the audio recordings, taala cycle annotations and notations in a machine readable format. <em>*This new 1.1 version includes additional information to align the notation w.r.t time.</em> <strong>Audio music content</strong> They feature 7 varnams in 7 rāgas sung by 5 young professional singers who received training for more than 15 years. They are all set to Adi taala. Measuring the intonation variations require absolutely clean pitch contours. For this, all the varṇaṁs are recorded without accompanying instruments, except the drone. <strong>Taala annotations</strong> The recordings are annotated with taala cycles, each annotation marking the starting of a cycle. We have later automatically divided each cycle into 8 equal parts. The annotations are made available as sonic visualizer annotation layers. Each annotation is of the format m.n where m is the cycle number and n is the division within the cycle. All m.1 annotations are manually done, whereas m.[2-8] are automatically labelled. <strong>Notations</strong> The notations for 7 varnams are procured from an archive curated by Shivkumar, in word document format. They are manually converted to a machine readable format (yaml). Each file is essentially a dictionary with section names of the composition as keys. Each section is represented as a list of cycles. Each cycle in turn has a list of divisions. <strong>Notations</strong> The notation is given a single time per section, however, to align the svaras with the tala annotations, structure information is given. The structure is given in yaml format, specifying the order of the sections, and how many svaras are sung per each tala tick. Broadly, there are just two only cases, 2 svaras per tick, and 4 svaras per tick.<br> The structure information has been added in the 1.1 version of the dataset. No code is given to load the structure information and relate it with the rest of data in the dataset. You may refer to the mirdata loader of the Carnatic Varnam dataset where in this case, tools to easily load the dataset are given. The structure annotations and the mirdata loader have been curated and implemented by Adithi Shankar and Genís Plaja. <strong>Possible uses of the dataset</strong> The distinct advantage of this dataset is the free availability of the audio content. Along with the annotations, it can be used for melodic analyses: characterizing intonation, motif discovery and tonic identification. The availability of a machine readable notation files allows the dataset to be used for audio-score alignment. Using this dataset If you use this dataset in a publication, please cite: Koduri, G. K., Ishwar, V., Serrà, J., &amp; Serra, X. (2014). Intonation analysis of rāgas in Carnatic music. Journal of New Music Research, 43(01), 73–94. http://hdl.handle.net/10230/25676 We are interested in knowing if you find our datasets useful! If you use our dataset please email us at [e-mail] and tell us about your research. <strong>Contact</strong> If you have any questions or comments about the dataset, please feel free to write to us. <br> Music Technology Group,<br> Universitat Pompeu Fabra,<br> Barcelona, Spain<br> mtg-info &lt;at&gt; upf &lt;dot&gt;edu http://compmusic.upf.edu/carnatic-varnam-dataset卡纳提克瓦尔纳姆数据集（Carnatic varnam dataset）收录了28段独唱人声录音，用于我们针对卡纳提克拉格（Carnatic raaga）的音高分析研究。本数据集包含音频录音、塔拉周期标注（taala cycle annotations）以及机器可读格式的乐谱。*本1.1版本新增了用于实现乐谱与时间轴对齐的额外信息。**音频内容** 本数据集包含7首基于7种拉格的瓦尔纳姆（varnam），由5名拥有15年以上专业训练的青年职业歌手演唱，全部采用阿迪塔拉（Adi taala）节拍结构。为精准测量音高变化，需要获取纯净无干扰的音高轮廓。为此，所有瓦尔纳姆均仅保留持续低音（drone）伴奏，未添加其他伴奏乐器。**塔拉周期标注** 录音已完成塔拉周期标注，每个标注均标记一个周期的起始点。后续我们将每个周期自动划分为8个均等分段。标注以声波可视化器标注图层的形式提供，所有标注采用`m.n`格式：其中`m`为周期序号，`n`为周期内的分段序号。所有`m.1`标注均为人工完成，而`m.[2-8]`标注则为自动生成。**乐谱标注** 7首瓦尔纳姆的乐谱源自Shivkumar整理的存档，初始格式为Word文档。经人工转换后，乐谱以机器可读的YAML（yaml）格式存储。每个乐谱文件本质为一个字典，以作品的段落名称作为键；每个段落以周期列表的形式表示，而每个周期又包含分段列表。**乐谱标注** 每个段落仅标注一次乐谱，但为了将斯瓦拉（svaras）与塔拉标注对齐，数据集额外提供了结构信息。该结构信息以YAML格式给出，明确了段落的顺序以及每个塔拉节拍（tala tick）所演唱的斯瓦拉数量。总体仅存在两种情况：每个节拍对应2个斯瓦拉，或每个节拍对应4个斯瓦拉。 本数据集1.1版本新增了上述结构信息，但未提供用于加载该结构信息并关联数据集其余内容的代码。您可参考卡纳提克瓦尔纳姆数据集的mirdata加载器（mirdata loader），其中包含了便捷加载数据集的工具。结构标注与mirdata加载器由Adithi Shankar与Genís Plaja整理并实现。**数据集可用场景** 本数据集的显著优势在于音频素材可免费获取。结合标注信息，该数据集可用于旋律分析：包括音高特征刻画、动机发现以及主音识别。机器可读格式的乐谱文件支持该数据集被用于音频-乐谱对齐任务。**数据集引用方式** 若您在学术成果中使用本数据集，请引用以下文献：Koduri, G. K., Ishwar, V., Serrà, J., & Serra, X. (2014). Intonation analysis of rāgas in Carnatic music. Journal of New Music Research, 43(01), 73–94. http://hdl.handle.net/10230/25676。我们非常期待了解您是否认为本数据集具备实用价值！若您使用了本数据集，请发送邮件至[e-mail]告知我们您的研究方向。**联系方式** 若您对本数据集有任何疑问或建议，欢迎随时联系我们。 音乐技术组（Music Technology Group） 庞培法布拉大学（Universitat Pompeu Fabra） 西班牙巴塞罗那 邮箱：[e-mail] 数据集主页：http://compmusic.upf.edu/carnatic-varnam-dataset