立命馆呼喊语料库 (RISC)

立命馆呼喊语料库（RISC）是由日本立命馆大学开发的一个包含多种呼喊语音样本的数据集，旨在支持紧急情况下的语音识别研究。该数据集通过录音实验收集了愤怒的呼喊、尖叫和欢呼等样本，并通过众包服务为每个呼喊语音样本分配了呼喊强度评级。RISC不仅用于语音类型分类任务，还用于呼喊强度预测任务，旨在提高安全系统在紧急情况下的识别能力。数据集的创建过程包括定义可能的呼喊语音句子列表，邀请实验参与者根据特定情境发出语音，并通过众包进行听觉实验以确定呼喊强度。RISC的应用领域主要集中在音频监控和监控系统中，以解决在紧急情况下快速准确识别呼喊语音的问题。

The Ritsumeikan Outcry Speech Corpus (RISC) is a dataset containing diverse outcry speech samples developed by Ritsumeikan University in Japan, aimed at supporting speech recognition research in emergency scenarios. This dataset collects samples such as angry outcries, screams and cheers through recording experiments, and assigns outcry intensity ratings to each outcry speech sample via crowdsourcing services. RISC is utilized for both speech type classification tasks and outcry intensity prediction tasks, with the goal of improving the recognition capability of security systems in emergency situations. The dataset creation process involves defining a list of potential outcry speech sentences, inviting experimental participants to generate speech based on specific scenarios, and conducting auditory experiments through crowdsourcing to determine outcry intensity. The application domains of RISC mainly focus on audio surveillance and monitoring systems, to solve the problem of rapidly and accurately recognizing outcry speech in emergency situations.