Dataset · seanghay
Khmer speech corpus
seanghay/km-speech-corpus
Around 10.4 hours of audio and text pairs in Khmer for training ASR and TTS models.
What's inside
Contains short speech audio recordings with text transcripts in Khmer, ranging from 10 to 100 thousand examples with a total duration of 10.4 hours. The recordings have a 16 kHz sample rate, and clip lengths range from 0.45 to 19.4 seconds with an average of about 2.5 seconds.
What you can do with it
- Training and fine-tuning automatic speech recognition (ASR) models for Khmer
- Training text-to-speech (TTS) systems in a low-resource language
- Evaluating existing multilingual audio models on Khmer
- Tasks
- Automatic speech recognition, Text → speech
- Languages
- Khmer
- Size
- 10K–100K rows
- Downloads this month
- 334
- Likes
- 10
Models trained on it
- Thareah/Tha-ASRAutomatic speech recognition 2.5K
- seanghay/whisper-small-khmer-v2Automatic speech recognition 1.3K
- PhanithLIM/whisper-small-aug-28-april-lightning-v1Automatic speech recognition 171
- PhanithLIM/whisper-large-v3-turbo-e3 118
- mlx-community/whisper-small-khmer-mlx-fp16Automatic speech recognition 108
- mlx-community/whisper-tiny-khmer-mlx-fp32Automatic speech recognition 74
- mlx-community/whisper-small-khmer-mlx-fp32Automatic speech recognition 68
- mlx-community/whisper-base-khmer-mlx-fp16Automatic speech recognition 65
How to load it
from datasets import load_dataset
ds = load_dataset("seanghay/km-speech-corpus")