Offline navigation customization and implementation method and device

Abstract

The invention discloses an off-line navigation customization and implementation method and device. The off-line navigation customization method comprises the steps that in response to offline navigation skills developed by a developer, built-in first speech recognition model and first semantic comprehension model are loaded; new functions added by the developer and dialogue logics customized by the developer are obtained; in response to the offline navigation skills issued by the developer, a second speech recognition model and a second semantic comprehension model which are associated with the new functions are trained, a dialogue model associated with the dialogue logics is trained, and the first speech recognition model, the first semantic comprehension model, the second speech recognition model, the second semantic comprehension model and the dialogue model are jointly generated into an off-line navigation package. The off-line navigation customization and implementation method anddevice can achieve a whole set of off-line navigation speech dialogue functions and support the personalized customization of the developer.

Description

technical field [0001] The invention belongs to the technical field of offline navigation, and in particular relates to a method and device for customization and realization of offline navigation. Background technique [0002] Among the related technologies, only a few companies have the basic voice technology required to provide offline navigation to the outside world. Some companies only provide one basic technology of offline speech synthesis, and then the navigation company integrates these two basic technologies into their own APP to realize the function of offline navigation voice broadcast navigation itinerary. Other companies provide a conversational artificial intelligence system with offline navigation dialogue, and then car manufacturers integrate the conversational artificial intelligence system into their own car machines to realize the function of offline navigation voice dialogue. [0003] In the process of realizing the present application, the inventor foun...

AI Technical Summary

Problems solved by technology

[0004] 1. One-way output: It can only provide speech synthesis and broadcasting capabilities when offline, and it is a one-way voice information output from the device to the person, and the person cannot input the voice information to the device

When people need to change functions during offline navigation, they still need to input information by hand, which is extremely unsafe during driving

[0005] 2. Poor integration ability: the offline delivery and integration method is used, and the integration and debugging costs are very high

Only suitable for large enterprises, not suitable for small and medium enterprises and individual developers

[0006] 3. The accuracy and customization of recognition semantics are not compatible: since the POI (Point of interest) involved in offline navigation is on the order of tens of millions, offline recognition requires a large amount of corpus training to produce, so offline recognition The update cost and cycle are particularly long, and customization is generally not accepted

[0007] 4. Poor customization of dialogue: Since it is integrated and delivered offline, the function of the entire offline navigation dialogue is realized by the developer's coding. It takes a lot of energy and time to open up, maintain and update this complete closed loop

Images