LAPS-Diff Hindi SVS Dataset
收藏资源简介:
LAPS-Diff Hindi SVS 数据集是一个由印度理工学院孟买分校电子工程系的研究团队创建的,包含约一小时时长的宝莱坞风格印地语歌曲数据集。该数据集包含37首歌曲,每首歌曲均由一位男性歌手演唱,共被分割成397个唱段。数据集旨在解决低资源情况下捕捉声乐风格、特定风格音高变化和语言相关特征的问题,并支持低资源情境下的声乐合成研究。数据集采用与 Opencpop 数据集相同的音乐乐谱格式,包含文本形式的歌词内容、对应的音素序列、关联的音符、音符时长、音素时长和连音信息。该数据集是首个针对声乐合成任务构建的印度音乐数据集。
The LAPS-Diff Hindi SVS Dataset was developed by a research team from the Department of Electrical Engineering, Indian Institute of Technology Bombay. It is a curated dataset containing approximately one hour of Bollywood-style Hindi songs, consisting of 37 tracks all performed by a single male vocalist, with the full content segmented into 397 vocal segments. This dataset is designed to address the core challenges of capturing vocal styles, style-specific pitch contours and language-related features in low-resource scenarios, and supports vocal synthesis research conducted under such low-resource conditions. It adopts the same musical score format as the Opencpop dataset, including textual lyric content, corresponding phoneme sequences, associated musical notes, note durations, phoneme durations, and inter-phoneme connection information. This is the first Indian music dataset specifically constructed for vocal synthesis tasks.

- 1LAPS-Diff: A Diffusion-Based Framework for Singing Voice Synthesis With Language Aware Prosody-Style Guided Learning印度理工学院孟买分校电子工程系 · 2025年



