efficient-speech

/

lite-whisper-tiny-fast

like 1

Follow

Efficient Speech 26

This PR enhances the model card for efficient-speech/lite-whisper-tiny-fast by:

1. Adding a top-level title: Using the paper's title # LiteASR: Efficient Automatic Speech Recognition with Low-Rank Approximation for better structure and clarity.
2. Including the paper abstract: Providing essential context about the model's design and findings directly from the paper.
3. Adding a "Quick Start" section with sample usage: Incorporating a Python code snippet from the GitHub README to demonstrate how to use the model with the transformers library. The processor in the example has been updated to openai/whisper-tiny to match the base_model of this specific repository, ensuring consistency and correctness for users.
4. Updating the "Benchmark Results" table: Replacing the existing table with the more comprehensive version found in the LiteASR GitHub repository, which covers a wider range of compressed Whisper models and provides a fuller picture of the project's performance.

These improvements make the model card more informative and user-friendly.