---
license: mit
task_categories:
- automatic-speech-recognition
language:
- km
---

# Khmer ASR Phase 1 Cache

This repository contains cached training artifacts used for Khmer ASR backbone evaluation.

## Source Datasets

- seanghay/khmer_grkpp_speech
- seanghay/km-speech-corpus
- KrorngAI/fleurs_openslr42_mpwt

## Contents

- combined_dataset.pt

## Usage

```python
import torch

data = torch.load("combined_dataset.pt")