wlejon/brosoundml-data
收藏资源简介:
该数据集包含brosoundml项目的训练权重和打包数据工件,用于支持语音处理任务,如词性标注、发音词典、语音合成和唤醒词检测。具体包括POS标注器模型权重、美式英语发音词典二进制文件、Kokoro-82M TTS模型的转换权重和配置文件、语音包文件以及computer唤醒词模型。这些工件按子系统组织在相应目录中,并附带许可证和来源信息。数据集可通过Hugging Face获取或通过脚本下载,适用于brosoundml库的部署和运行。
This dataset contains the training weights and packaged data artifacts of the brosoundml project, which are designed to support speech processing tasks including part-of-speech tagging, pronunciation dictionary construction, speech synthesis, and wake word detection. Specifically, it includes the model weights of the POS tagger, binary files of the American English pronunciation dictionary, converted weights and configuration files of the Kokoro-82M TTS model, speech package files, and the computer wake word model. These artifacts are organized into corresponding directories by subsystem, and are accompanied by license and source information. The dataset can be obtained via Hugging Face or downloaded through scripts, and is suitable for the deployment and operation of the brosoundml library.




