ATC-EnRoute-Synthetic-Dataset
收藏资源简介:
ATC En-Route Synthetic Dataset 是一个合成的空中交通管制通信数据集,专为航路(巡航)阶段的训练和评估而设计。数据集包含初始联系和操作指令(如航向、速度、高度变化),总时长约为3小时25分钟。数据以同步的立体声音频轨道和CSV元数据格式提供,其中元数据包括文本转录、说话者、目标呼号(target_callsign)和意图(intent)等字段。该数据集的生成源于学士论文《基于语音识别和机器学习的ATC人工智能助手开发》,适用于音频分类和自动语音识别等任务。
The ATC En-Route Synthetic Dataset is a synthetic air traffic control communication dataset designed for training and evaluation during the en-route (cruise) phase. It includes initial contacts and operational instructions (such as heading, speed, and altitude changes), with a total duration of approximately 3 hours and 25 minutes. The data is provided in synchronized stereo audio tracks and CSV metadata format, with metadata fields including text transcription, speaker, target callsign, and intent. This dataset was generated from the bachelors thesis Development of an ATC AI Assistant Based on Speech Recognition and Machine Learning and is suitable for tasks such as audio classification and automatic speech recognition.
数据集概述
ATC En-Route Synthetic Dataset 是一个合成的空中交通管制(ATC)通信数据集,专为航路(巡航)阶段的训练和评估而设计。
基本信息
- 许可证: MIT
- 任务类型: 音频分类、自动语音识别
- 总时长: 约3小时25分钟
数据内容
- 通信场景: 初始联系及操作指令(航向、速度、高度变化)
- 数据格式: 同步立体声音频轨道及CSV元数据
数据特征
数据集包含以下字段:
- text: 文本内容
- speaker: 说话人
- target_callsign: 目标呼号
- intent: 意图
用途
该数据集用于学士论文《基于语音识别和机器学习的ATC人工智能助手开发》。
生成代码
生成该数据集的代码可在GitHub上获取:https://github.com/Joseagl14/ATC-Synthetic-Data-Generator




