Publish time: 2022-03-28 00:00:00
05T-1110325
Know-how
This Mandarin Chinese speech recognition system is developed using Kaldi toolkit. The core modules include Chinese pronunciation dictionary, acoustic model, language model and recognition system architecture. The system includes a real-time speech recognition core and an offline speech recognition core. The former can be applied to various application systems that require real-time recognition of input speech, such as human-computer interaction systems and smart speakers. The latter can be used to convert audio recordings to verbatim transcripts or to subtitle a video.
- Maturity: After years of hard work, the Mandarin Chinese speech recognition technology has become quite mature, and the functions of the Mandarin Chinese speech recognition system developed are quite complete.
- Easy to use: The recognition core can be easily connected to downstream application systems.
- various application systems that require real-time recognition of input speech, such as human-computer interaction systems and smart speakers
- transcription of voice recordings
- spoken document retrieval

Figure1 The demonstration interface for Mandarin Chinese speech recognition. It currently provides four functions: real-time speech recognition, uploading audio files for recognition, uploading YouTube video links for recognition, and using API for recognition
Hsin-Min Wang; Ming-Tat Ko; Yu Tsao
If you are interested in this technology, please fill in the information on the webpage under the manufacturer option (Request for Technology Licensing, Material Transfer, or Sponsored Research) on the website of this office

Mandarin Chinese Speech Recognition System