Introduction and Application of Far-field Speech Recognition Technology

With the development of technology, far-field speech recognition has become one of the most successful technologies in the global consumer electronics field with the rise of smart speakers.

Far-field speech recognition is an important technology in the field of speech interaction, mainly facing echo interference, indoor reverberation, multi-signal source interference. This technology has important applications in scenarios such as smart home (smart speaker, smart TV, etc.), meeting rooms, vehicle navigation.

Far-field speech recognition needs to be completed through the combination of front-end and back-end. In the language front-end processing module, it is mainly solved by introducing microphone array. Microphone array technology refers to a system composed of a certain number of acoustic sensors (usually microphones) used to sample and process the spatial characteristics of sound field. Using the microphone array to distinguish the direction of the sound source, realize the sound source localization, realize the extraction and separation of the sound source signal, enhance the voice signal, and realize the reverberation effect at the same time. Combined with AI algorithm, that is, the background recognition engine, through the complex combination of software and hardware, far-field voice (FFV) can realize human-computer interaction (HMI), allowing users to "talk" with the smart box at a distance.

What applications is far-field speech recognition suitable for?

1. Smart Home

TV/Set-top box: Rich media content makes its operation complicated, and also limited by the inefficient input method of the remote control, making TV the most suitable device for voice transformation.

Speakers: Speakers are the first to detonate the market because of their low cost (no screen and video resources required).

2. Vehicle Navigation

The unique advantages of in-vehicle voice technology-help drivers reduce their dependence on the operation of in-vehicle equipment and increase driving safety. The wind noise is loud when the window is opened. By using far-field speech recognition to give voice commands such as answering calls, broadcasting music, and route guidance, driving is safer and you can focus more on road conditions.

3. Conference transcription

Conferencing is one of the most valuable and challenging applications for far-field speech recognition technology. This applications involves complex acoustic conditions, with challenges such as overlapping speech, unknown number of speakers, far-field signals in large conference rooms, noise, and reverberation.

Smart far-field voice is a function that combines AI voice search to change the user experience. It allows you to speak commands at a long distance (usually 1m-10m), and the machine will give feedback after recognizing the human voice.

KA1 Smart speaker realizes far-field voice operation by connecting to Google Assistant. It adopts a dual-microphone array and picks up sound at a distance of 3 to 5 meters, which can effectively reduce the noise in the surrounding environment. This is a far-field voice interaction product launched for different intelligent application scenarios.