Daniel ASR Notes
Daniel's working notes on ASR models: quirks, limits and findings
About This Project
This repository serves as a technical knowledge base for documenting the behavior, constraints, and operational quirks of modern speech recognition models. It catalogs findings gathered through rigorous fine-tuning processes and consistent real-world dictation testing, providing a structured look at model performance beyond standard benchmarks.
Each documentation entry is maintained as a Markdown file, which is automatically rendered into a portable PDF format using Pandoc and WeasyPrint. The project is designed for researchers and developers working with speech-to-text technologies who require specific insights into model edge cases and character support limitations.