Vedavani
收藏资源简介:
Vedavani是一个用于梵文吠陀诗歌自动语音识别(ASR)的基准语料库。该数据集由来自《梨俱吠陀》和《阿闼婆吠陀》的30,779个标记音频样本组成,总时长为54小时。这个数据集捕捉了梵文独特的音调和节奏特征,为研究和开发梵文诗歌的ASR系统提供了宝贵资源。数据集的创建涉及从维基百科获取文本资源,从互联网档案馆获取音频转录,并手动对齐文本和音频数据。Vedavani数据集旨在解决梵文诗歌的自动语音识别问题,为研究梵文语言提供了新的视角和方法。
Vedavani is a benchmark corpus for automatic speech recognition (ASR) of Sanskrit Vedic poetry. This dataset consists of 30,779 labeled audio samples sourced from the Rigveda and Atharvaveda, with a total duration of 54 hours. It captures the unique tonal and rhythmic characteristics of Sanskrit, serving as a valuable resource for research and development of ASR systems targeting Sanskrit poetry. The creation of the Vedavani dataset involved acquiring text resources from Wikipedia, obtaining audio transcriptions from the Internet Archive, and manually aligning the text and audio data. The Vedavani dataset aims to address the automatic speech recognition task for Sanskrit poetry, providing new perspectives and methodologies for research on the Sanskrit language.
数据集概述
基本信息
- 数据集名称:Vedavani
- 托管平台:GitHub
- 托管地址:https://github.com/SujeetNlp/Vedavani
当前状态
- 数据集尚未公开
- 代码和数据集即将发布
备注
- 无其他可用信息

- 1Vedavani: A Benchmark Corpus for ASR on Vedic Sanskrit Poetry印度理工学院,加尔各答分校 · 2025年



