Dictation Corrector

  • Redis
  • OpenAI Whisper
  • Python
  • TypeScript
  • React

This Chrome extension enhances the accuracy of proper noun recognition in voice dictation services. Leveraging foundational models locally and on cloud, it detects and corrects proper nouns that may be misinterpreted during voice-to-text transcription. When a misinterpretation is identified, the extension automatically adjusts the term and stores the correction for future reference, enabling consistent accuracy and personalized adaptation over time.

The core functionality involves real-time monitoring of dictation inputs, with each detected proper noun processed through a foundation model to ensure correctness. Local storage is utilized to track all corrections, allowing the extension to "learn" from each adjustment and prevent circular correction by the model, creating a tailored experience for each user. This hybrid approach—combining cloud-based processing with local memory—offers both performance efficiency and adaptability, ensuring a responsive transcription experience that continuously improves with use.

Designed for seamless integration into various voice-to-text workflows, the extension enhances reliability by reducing common transcription errors around proper nouns.

2024-present © akjadhav

inspired from ashish