Double voice recognition method of service robot sharing one set of microphone array

Abstract

The invention relates to a double voice recognition method of a service robot sharing one set of microphone array. The double voice recognition method is characterized in that one set of microphone array and two sets of voice recognition systems are installed on a body of the service robot, the microphone array is an annular microphone array or a linear microphone array, and the two sets of voice recognition systems are an off-line voice recognition system and an online voice recognition system. The operating method is that the microphone array sends acquired voice information to the two sets of voice recognition systems simultaneously, the two sets of voice recognition systems are enabled to perform voice recognition respectively and convert the voice into texts, and corresponding operations are executed according to matching conditions of the content of the respectively acquired texts and an off-line voice text library which includes daily conversations, command words and smart home control instructions; if the acquired texts acquired by recognition of the off-line voice recognition system are matched with the off-line voice text library, off-line voice recognition is utilized to perform interaction; and if the texts acquired by recognition of the off-line voice recognition system are not matched with the off-line voice text library, online voice recognition is utilized to perform interaction.

Description

technical field [0001] The invention relates to a dual-speech recognition method for a service robot sharing a set of microphone arrays. It is characterized by installing a microphone array and two voice recognition systems on the service robot body, wherein the microphone array is a circular microphone array or a linear microphone array, and the two voice recognition systems are an offline voice recognition system and an online voice recognition system. Its working method is that the microphone array sends the obtained voice information to two sets of voice recognition systems at the same time, so that the two sets of voice recognition systems can perform voice recognition and convert it into text respectively, and the text content obtained by each can be compared with the offline voice text library (daily dialogue, Command words, smart home control instructions) match, then perform the corresponding operation: when the text recognized by the offline speech recognition system...

AI Technical Summary

Problems solved by technology

[0006] At present, a set of voice recognition system is generally used on the service robot body, and it is mainly based on online voice recognition, which requires a good network environment support. If the service robot has a network connection failure or the bandwidth of the online voice cloud outlet is affected, voice will appear. Identify system failures and delays that affect the effect of human-computer interaction

Although some speech recognition solution companies are aware of this problem and have assisted the offline recognition function, that is, when the online speech recognition cannot provide online speech cloud recognition services, they switch to the offline speech recognition mode, but because the offline speech recognition is not optimized according to the scene , the effect of offline recognition is poor. This solution of only one speech recognition system at the same time online and then offline does not meet the actual needs well, and a more suitable combination of online speech recognition and offline speech recognition is needed.

Images