Japanese Vowel Speech Dataset

该数据集包含了60个音频文件，其中一位母语为日语的女性朗读了30个人工合成的句子，朗读速度自然。这些句子特别设计，包含了特定的单词和音素组合，以确保数据集中包含了日本语五个元音形成的五个单词。此外，这些数据被编码成了12维的梅尔频率倒谱系数（MFCC）时间序列数据。该数据集的规模为60个音频文件，其任务旨在进行单词和音素的发现研究。

This dataset contains 60 audio files, generated by a native Japanese female speaker reciting 30 artificially synthesized sentences at a natural speaking rate. These sentences are specially engineered to contain specific word and phoneme combinations, ensuring coverage of five words formed by the five vowels of the Japanese language. Additionally, the audio data has been encoded into 12-dimensional Mel-Frequency Cepstral Coefficients (MFCC) time-series data. Designed for research on word and phoneme discovery, this dataset consists of exactly 60 audio files.