美式英语自然对话手机采集语音数据集

美式自然对话的应用场景：随着语音识别技术在智能客服、智能会议等更多自然场景下的落地，说话人的发音习惯更贴近日常沟通行为，在发声时会有大量的连音、吞音、发音变形、咬字不清等。就迫切需要自然对话方式的语音数据。本数据可用于美式英语语音识别模型的训练和测试，提升模型在自然对话方式下的识别率。

Application Scenarios of American Natural Dialogue: As automatic speech recognition (ASR) technologies are being deployed in more natural scenarios such as intelligent customer service and smart meetings, the pronunciation habits of speakers have become closer to those in daily communication. During speech production, there are numerous phenomena including connected speech, elision, phonetic distortion and unclear articulation, which creates an urgent demand for speech data in natural dialogue scenarios. This dataset can be used for training and testing American English speech recognition models to improve the models' recognition accuracy under natural dialogue scenarios.