Speech Recognition TRIZ Cases
- Hybrid Target Word Spotting for Fast Automatic Speech Recognition
A hybrid ASR approach spots context-driven target words without quadratic decoding delays.
- Pairwise UWB Earbud Ranging for Assistant Target Arbitration
Pairwise UWB multilateration identifies the intended device before speech recognition, reducing wasted processing and power.
- Automated Speech Recognition Correction with Iterative Hint Words
This ASR correction process submits related hint words repeatedly to improve transcription accuracy without changing the core recognizer.
- Adaptive Text-to-Speech With Feedback-Based Contextual Speech Patterns
Natural language processing categorizes context and user feedback to refine individualized speech patterns for nuanced conversations.
- TWS Voice Recognition Engine Sharing Under Memory Limits
This case distributes voice recognition engines between TWS and external devices, expanding service variety without local memory overload.
- RNN-T Audio Training with Gradient-Based Pruning for Lower Memory
This RNN-T approach prunes joint-training results to reduce memory and computation while accelerating audio model training.
- Starting-Word Speaker Recognition for Personalized Display Results
An NLP-linked display recognizes speakers only after selected starting words, balancing personalized results with processing overhead.