Speech Correction System Using Preliminary Voice Sampling for Accent Adaptation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing speech recognition models face recognition errors due to differences in accents, leading to incorrect interpretation of voice commands, which hampers user interaction with voice-based systems.
Innovation Solution
A speech correction system and method that includes a storage device and processing device with an audio receiver, speech recognition engine, calculation module, and determination module, which receives voice inputs, generates candidate vocabularies and their probabilities, performs operations on these probabilities, and determines results greater than a score threshold to improve recognition accuracy by storing output results in a database.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a general speech recognition model is used, then the system can process various voice inputs, but recognition accuracy deteriorates due to accent differences
Solution Approach 1:
The system performs preliminary actions by collecting multiple voice inputs from the same user before formal recognition, and pre-processes these inputs to generate candidate vocabularies and probabilities. This preliminary processing creates a customized vocabulary set specific to each user's accent patterns, which is then used to improve subsequent recognition accuracy without sacrificing the system's ability to handle diverse voice inputs.
Solution Approach 2:
The system applies local quality by creating user-specific vocabulary sets and probability distributions tailored to individual accent characteristics. Instead of using a uniform recognition model for all users, the system generates localized vocabulary adjustments based on each user's voice samples, thereby improving recognition accuracy for that specific user while maintaining overall system versatility.
2Measurement precision
If multiple voice inputs are collected and processed, then recognition accuracy improves, but processing time increases
Solution Approach 1:
The system performs voice collection and preliminary processing as a one-time setup action before actual recognition tasks. By completing the vocabulary generation and probability calculation in advance, the system avoids repeated processing during normal operation, thus improving recognition accuracy while minimizing time loss during actual use.
Solution Approach 2:
The system performs self-service by automatically collecting voice samples and generating customized vocabulary sets without requiring manual intervention or repeated processing. Once the initial voice inputs are collected, the system autonomously processes this data to create user-specific recognition parameters, eliminating the need for continuous time-consuming adjustments.
3Measurement precision
If customized vocabularies are generated for different users, then recognition precision improves, but system complexity increases
Solution Approach 1:
The system generates customized vocabularies and probability distributions as preliminary data structures that are stored and reused. This preliminary generation approach avoids the need for complex real-time calculations during recognition, as the customized parameters are pre-computed and readily available for matching against new voice inputs.
Solution Approach 2:
The system uses copying by creating user-specific copies of the vocabulary structure from a general vocabulary framework. Instead of inventing entirely new complex structures for each user, the system copies the basic vocabulary framework and adjusts it with user-specific probability values and candidate words, thereby reducing system complexity while maintaining customization benefits.
Data Source
AI summary
The speech correction system includes a storage device and a processing device. The storage device stores a first database. The processing device includes an audio receiver, a speech recognition engine, a calculation module, and a determination module. The audio receiver receives multiple voice inputs. The speech recognition engine recognizes the voice inputs, generates multiple candidate vocabularies corresponding to each of the voice inputs, and generates a vocabulary probability corresponding to each of the candidate vocabularies. The calculation module performs a specific operation on the vocabulary probabilities corresponding to the same candidate vocabulary, to generate a plurality of corresponding operation results. The determination module determines whether each of the operation results is greater than a score threshold, and stores at least one output result that is greater than the score threshold to the first database.