Speech Commands

Speech Commands数据集由谷歌大脑创建，旨在帮助训练和评估关键词识别系统。该数据集包含105,829条语音命令，涉及35个常用单词，每条语音命令存储为1秒的WAVE格式文件。数据集通过开放源代码的网络应用程序收集，使用WebAudioAPI录制，确保了数据的多样性和实用性。该数据集主要用于训练小型模型，以检测单一单词的语音命令，适用于物联网或机器人应用等场景，旨在解决语音交互中的关键词识别问题。

The Speech Commands Dataset was developed by Google Brain to aid in the training and evaluation of keyword recognition systems. It comprises 105,829 speech commands spanning 35 common words, with each command stored as a 1-second WAVE format file. The dataset was collected through an open-source web application, with recordings made using the WebAudioAPI, ensuring the diversity and practicality of the collected data. This dataset is mainly used to train small-scale models for detecting single-word speech commands, and is suitable for scenarios such as Internet of Things (IoT) or robotics applications, with the goal of resolving keyword recognition issues in speech interaction.