谷歌旗下DeepMind在語音生成領域取得突破

Google’s DeepMind division has claimed a big leap forward in producing computer-generated speech, potentially bringing forward the day when machines sound as natural as humans.

谷歌(Google)旗下的DeepMind稱，在計算機語音生成領域取得巨大飛躍。這一技術進步可能會令機器像人一樣自然發聲的一天提前到來。

Researchers at the UK-based artificial intelligence arm said tests of their system with human listeners, though unavoidably subjective, had shown that the technology halves the quality gap that exists between today’s computer systems and human speech.

這家總部駐英國的人工智能分公司的研究人員表示，由人類聽衆對其系統的測試（儘管這種測試不可避免地帶有主觀性）顯示，該技術將目前計算機系統與人類說話間存在的質量差距縮小了一半。

One person familiar with the research said the system, called WaveNet, was not being employed in any of Google’s products yet, and that the massive computing power it needed to generate speech made it impractical for real world use for the foreseeable future.

一位知情人士表示，這一名叫WaveNet的系統尚未被用於谷歌任何產品。該技術在生成語音時需要極大的運算能力，令它在可以預期的未來無法實際運用到真實世界。

The research that led to WaveNet, along with a number of samples of speech generated by the system, was published online on Thursday.

週四，WaveNet的相關研究以及一系列該系統生成的語音樣本被髮布到網上。