Voice Prompt Caching Reduces Power and Network Usage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wireless devices face challenges in providing natural-sounding and detailed voice prompts due to the limitations of pre-recorded speech data, which consume significant processing resources and power when using text-to-speech (TTS) conversion, and rely heavily on network access for each conversion request.
Innovation Solution
Implementing a system where an electronic device stores synthesized speech data locally after an initial TTS conversion request to a server, allowing for subsequent reuse of this data instead of re-requesting conversion for identical prompts, and using pre-recorded data when network access is unavailable or conversion is delayed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If TTS conversion is performed at the electronic device, then detailed and natural-sounding voice prompts are achieved, but processing resources and power consumption increase significantly
Solution Approach 1:
The patent extracts the TTS conversion function from the electronic device and relocates it to a remote server. The electronic device sends text prompts to the server, which performs the conversion and returns synthesized speech data. This extraction eliminates the need for the electronic device to have TTS conversion capabilities, significantly reducing its processing requirements and power consumption while still delivering detailed and natural-sounding voice prompts.
Solution Approach 2:
The patent introduces a remote server as an intermediary between the electronic device and the voice prompt output. The server acts as a mediator that receives text prompts, performs TTS conversion using advanced algorithms, and returns the synthesized speech data. This intermediary approach allows the electronic device to leverage the server's processing power without bearing the energy cost of performing conversion locally.
2Device complexity
If TTS conversion is offloaded to an external server, then processing resources are reduced, but network access is required for each conversion request
Solution Approach 1:
The patent implements preliminary action by caching synthesized speech data locally in the electronic device's memory after initial TTS conversion requests. When the same or similar text prompts are encountered again, the device checks its local cache first and retrieves stored speech data without needing to access the network. This preliminary caching action reduces network dependency and allows the device to operate independently when cached data is available.
3Loss of energy
If pre-recorded speech data is used, then network access is minimized, but detailed and natural-sounding voice prompts cannot be provided
Solution Approach 1:
The patent uses preliminary action by pre-converting text prompts to synthesized speech data via TTS conversion and storing them in the electronic device's memory before they are needed. This advance preparation creates a local cache of speech data that can be retrieved without network access. When the device needs to provide voice prompts, it first checks the cache for pre-converted data, ensuring both low network usage and access to detailed, natural-sounding prompts.
Solution Approach 2:
The patent implements dynamics by creating a hybrid system that adaptively switches between different data sources. The device dynamically determines whether to use pre-recorded speech data, cached TTS converted data, or request new TTS conversion based on availability and requirements. This dynamic approach allows the system to optimize between network usage and prompt quality in different operational scenarios.
Data Source
AI summary
An electronic device includes a processor and a memory coupled to the processor. The memory stores instructions that, when executed by the processor, cause the processor to perform operations including determining whether a text prompt received from a wireless device corresponds to first synthesized speech data stored at the memory. The operations include, in response to a determination that the text prompt does not correspond to the first synthesized speech data, determining whether a network is accessible. The operations include, in response to a determination that the network is accessible, sending a text-to-speech (TTS) conversion request to a server via the network. The operation further include, in response to receiving second synthesized speech data from the server, storing the second synthesized speech data at the memory.