CATSLU 中文音频文本口语理解数据集

CATSLU 是一个中文语音 +NLU 文本理解的对话数据集，从语音信号到理解端到端进行实验，例如直接从音素建模语言理解（而不是词或标记）。该数据集来自第一届中文音文本口语理解挑战赛，包括测试数据集和结果、训练和验证数据集、基线和手册。

CATSLU is a conversational dataset for Chinese speech and natural language understanding (NLU). It enables end-to-end experiments from speech signals to language understanding, such as modeling language understanding directly from phonemes rather than words or tokens. This dataset is derived from the First Chinese Phonetic-Text Spoken Language Understanding Challenge, and includes test datasets and their corresponding results, training and validation datasets, baseline models, and supporting documentation.

搜集汇总

背景与挑战

背景概述

CATSLU是一个中文语音与自然语言理解结合的对话数据集，用于从语音信号到理解端到端的实验，包含训练、验证和测试数据及基线结果。数据集来自第一届中文音文本口语理解挑战赛，大小为1.13 GB。