Whisper

Whisper 是一种自动语音识别 (ASR) 系统，使用从网络收集的 680,000 小时多语言和多任务监督数据进行训练。 我们表明，使用如此庞大且多样化的数据集可以提高对口音、背景噪音和技术语言的鲁棒性。 此外，它还支持多种语言的转录，以及将这些语言翻译成英语。 我们正在开源模型和推理代码，以作为构建有用应用程序和进一步研究稳健语音处理的基础。

Whisper is an automatic speech recognition (ASR) system trained on 680,000 hours of multilingual and multitask supervised data collected from the web. We demonstrate that training on such a large and diverse dataset improves robustness to accents, background noise, and technical jargon. Furthermore, it supports transcription in multiple languages, as well as translation of those languages into English. We are open-sourcing both the model and inference code to serve as a foundation for building useful applications and further research in robust speech processing.

搜集汇总

背景与挑战

背景概述

Whisper是一个基于68万小时多语言监督数据训练的自动语音识别系统，由OpenAI于2022年9月发布。该系统具备对口音和噪音的鲁棒性，支持多种语言的转录及翻译成英语，旨在为语音处理应用和研究提供基础。