Voice Recognition Exclusion Vocabulary and Temporal Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing voice recognition systems face challenges in distinguishing between intended voice commands and unintended voices, leading to false recognition, particularly in triggerless recognition scenarios where conversation or monologue voices are misinterpreted as command inputs.
Innovation Solution
A voice recognition processing apparatus that includes a recognition result determiner with an exclusion vocabulary rejecter, utterance duration determiner, and utterance form determiner, which analyzes voice data to differentiate between voice operations and non-operation voices by using exclusion lists, utterance duration, and utterance form information, in conjunction with a voice recognition server for enhanced accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If triggerless recognition is performed to increase convenience, then ease of operation is improved, but false recognition increases due to inability to distinguish between command voices and non-command voices
Solution Approach 1:
The system performs preliminary actions by measuring utterance duration and detecting utterance form (pause periods) before making the final recognition decision. This allows the system to filter out non-command voices based on their temporal characteristics before they can cause false recognition, while still maintaining triggerless operation for genuine commands.
Solution Approach 2:
The system changes the parameters used for voice recognition by incorporating temporal parameters (utterance duration, pause period lengths) in addition to traditional voice content analysis. This multi-parameter approach enables the system to distinguish between command voices and non-command voices while maintaining triggerless recognition capability.
2Ease of operation
If voice collection is always performed without limitation, then ease of operation is improved, but false recognition occurs due to inability to distinguish command voices from conversation or monologue
Solution Approach 1:
The system performs preliminary analysis of utterance duration and utterance form (detecting pause periods) before determining whether the voice should be recognized as a command. This preliminary action allows continuous voice collection while preventing false recognition of non-command voices through temporal characteristic analysis.
Solution Approach 2:
The system introduces intermediary parameters (utterance duration, pause periods) that act as mediators between the raw voice input and the final recognition decision. These intermediary measurements provide additional information to distinguish command voices from conversation or monologue without limiting continuous voice collection.
3Device complexity
If only vocabulary comparison is used for recognition, then device complexity is reduced, but recognition accuracy deteriorates due to false recognition of non-command voices
Solution Approach 1:
The system performs preliminary measurements of utterance duration and utterance form before vocabulary comparison. These preliminary actions add minimal complexity to the system while significantly improving recognition accuracy by filtering out non-command voices based on their temporal characteristics before they reach the vocabulary matching stage.
Solution Approach 2:
The system adds another dimension to voice recognition by incorporating temporal parameters (duration, pause periods) in addition to the traditional vocabulary dimension. This multi-dimensional approach improves recognition accuracy without substantially increasing device complexity, as the additional measurements are computationally simple.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Operativity of voice operation is improved. For this purpose, in voice recognition processing apparatus (100), voice acquirer (101) is configured to acquire a voice uttered by a user and to output voice information. Voice recognizer (102) is configured to convert the voice information into first information. Storage device (170) previously stores a dictionary in which an exclusion vocabulary is registered. Recognition result determiner (104) compares the first information with the exclusion vocabulary. When the first information includes a word that agrees with a word included in the exclusion vocabulary, recognition result determiner (104) determines that the first information is information to be rejected. When the first information does not include the word that agrees with a word included in the exclusion vocabulary, recognition result determiner (104) determines that the first information is information to be executed.