解读智能语音技术的2020：跨语种语音技术成高频关键词，商业化“加速度”落地

{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"},{"type":"strong"}],"text":"本文是 InfoQ“解读 2020”年终技术盘点系列文章之一。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"2016 年，深度学习和深度神经网络的突破使得智能语音识别的准确率第一次达到了人类水平，也促使智能语音技术进入到落地阶段。尤其是近几年，语音识别技术逐渐走向成熟，在万物互联趋势下，智能语音技术在教育、金融等各个行业落地日益深入。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"2020年，智能语音赛道的发展态势如何，有哪些重大的技术突破，在各行业有哪些应用和落地进展？2021年，这个赛道又蕴含着哪些发展机会？InfoQ 采访了网易有道AI语音团队的负责人孙艳庆，对智能语音领域过去一年的发展进行总结、回顾与探讨，并展望明年的发展趋势。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"领域涌现多项重大技术突破，挑战犹在"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"刚刚过去的2020年，智能语音领域出现了多个重要的技术创新与突破。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"“跨语种语音技术是智能语音技术在2020年的高频关键词。无论是语音识别还是语音合成，都有这样的形式”，孙艳庆表示。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"例如，谷歌在2019年提出了跨语种的TTS；苹果在IOS14中，推出了新版翻译功能，支持自动的语言和语音识别，无需用户手动预先选择语言。网易有道在今年9月上线的王源“明星语音”也借鉴了这个框架。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"**孙艳庆认为，Transformer在更多领域方向（语音\/图像）、以及建模技术的不断发展，使得其建模精度越来越强，也是2020的一个关键技术突破。**今年10月，网易有道上线了基于Transformer+CTC 架构的新一代有道ASR引擎，实践验证，语音识别准确率和用户体验得到了显著提升，大幅超越了线上采用主流算法的效果。"}]}]}