README.md

metadata

title: Phoneme Detection Leaderboard
emoji: 🎤
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: 4.44.1
app_file: app.py
pinned: false

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

Phoneme Detection Leaderboard

A clean, simplified phoneme detection leaderboard based on the open_asr_leaderboard interface.

Features

• Clean Interface: Uses the same interface structure as open_asr_leaderboard
• Phoneme Evaluation: Evaluates models on phoneme recognition tasks
• Multiple Datasets: Supports evaluation on multiple phoneme datasets
• Model Request System: Allows users to request evaluation of new models

Structure

├── app.py                 # Main Gradio application
├── constants.py          # Constants and text definitions
├── utils_display.py      # Display utilities and column definitions
├── init.py              # Initialization and hub integration
├── phoneme_eval.py      # Core phoneme evaluation logic
├── utils/               # Utility modules
│   ├── load_model.py    # Model loading and inference
│   ├── audio_process.py # Audio processing and PER calculation
│   └── cmu_process.py   # CMU to IPA conversion
├── requirements.txt     # Python dependencies
└── README.md           # This file

Usage

1. Install dependencies:

   pip install -r requirements.txt
   

2. Run the application:

   python app.py
   

3. Run evaluation:

   python phoneme_eval.py
   

Evaluation

The leaderboard evaluates models on:

• PER (Phoneme Error Rate): Lower is better
• Average Duration: Processing time per sample

Models are ranked by Average PER across all datasets.

Datasets

• phoneme_asr: General phoneme recognition dataset
• kids_phoneme_md: Children's speech phoneme dataset