TellWorldAI World Archive

Home›News›Release

ReleaseStatistical learning

China's National Engineering Laboratory for Speech and Language Information Processing is unveiled

On 27 September 2011 the National Engineering Laboratory for Speech and Language Information Processing was unveiled at USTC, jointly built with iFlytek to connect speech research, product testing and standards work.

Microphone and blank plaque in a speech lab.

What speech technology has to solve

Speech synthesis (turning written text into spoken audio) and speech recognition (turning recorded speech into text) look like opposite tasks, but they depend on the same scarce materials: large collections of recorded and transcribed speech, models trained on those collections, and hardware fast enough to run the models while a person is waiting for an answer. Natural language processing sits between the two, working out what the recognised words mean and how the sentence to be spoken should be built. Intelligent human–machine speech interaction is the combination of all three — a system a person can talk to, which answers out loud. Work of this kind is expensive, and it has to be tried on real products before anyone knows whether it holds up in noisy rooms and with different accents. That is the gap an engineering laboratory is meant to fill: it puts university research, company products and industry standards on the same chain.

The unveiling

On 27 September 2011 the National Engineering Laboratory for Speech and Language Information Processing was unveiled at the University of Science and Technology of China (USTC), according to a report in Science Times [1]. The laboratory had been approved by the National Development and Reform Commission [1]. The report called it the only national-level research platform in China's intelligent speech field, and the first national engineering laboratory in Hefei [1].

How it was organised

The laboratory would build 11 core research laboratories at USTC, covering speech synthesis, speech recognition, natural language processing and intelligent human–machine speech interaction [1]. The company iFlytek would host the technology research and development, design and development, test and verification, and major application demonstration platforms [1]. The laboratory was jointly built by USTC and iFlytek [2]. According to USTC's own news page, the first phase would work on nine core technologies and four engineering directions, including speech chips and standards work [2]; the laboratory would also help formulate national standards and take the lead in international standards [2].

That division of labour is the point of the arrangement. The university side supplies the research laboratories where the underlying methods are studied; the company side supplies the places where a method is turned into a product, tested, verified and demonstrated at scale. Standards work sits alongside both, because speech systems only become useful infrastructure when a recogniser, a synthesiser and the services around them can be fitted together by people who did not build them.

Where the two partners stood

At the time of the unveiling, USTC and iFlytek had taken back 80 percent of the Chinese-language mainstream speech market, the report said [1]. The two institutions had also won first place in the Blizzard Challenge English speech synthesis competition for six consecutive years [1]. One figure describes a position in the domestic market; the other describes a record in an international evaluation held in English. The laboratory was meant to turn both into something that could be produced repeatedly, rather than resting on the results of individual teams.

Why it is worth recording

A research direction usually asks for an engineering laboratory only once it has become mature enough that papers and demonstrations are no longer sufficient. Someone has to turn prototypes into systems that can be tested, verified and manufactured; someone has to write the standards that let different makers' systems meet. This laboratory put the university's research rooms and the company's testing and demonstration platforms into a single framework [1][2], and wrote standards work into its tasks from the first phase [2]. The report's description of it as the only national-level platform in the field, and the first of its kind in Hefei, marks the moment when speech technology in China was given its own engineering infrastructure rather than a single group's project [1].

Sources

  1. 1
  2. 2
    「语音」国字号 合肥拔「头筹」中国科学技术大学新闻网

Cite this page: China's National Engineering Laboratory for Speech and Language Information Processing is unveiled. TellWorld, https://www.tellworld.net/news/voice-language-engineering-lab-2011. Updated 2026-11-07

Related news

View all