|
--- |
|
license: mit |
|
datasets: |
|
- thennal/IMaSC |
|
language: |
|
- ml |
|
model-index: |
|
- name: Malwhisper-v1-medium - Kurian Benoy |
|
results: |
|
- task: |
|
type: automatic-speech-recognition |
|
name: Automatic Speech Recognition |
|
dataset: |
|
name: Common Voice 11.0 |
|
type: mozilla-foundation/common_voice_11_0 |
|
config: ml |
|
split: test |
|
args: ml |
|
metrics: |
|
- type: wer |
|
value: 23.69 |
|
name: WER |
|
- type: cer |
|
value: 12.17 |
|
name: CER |
|
library_name: transformers |
|
--- |
|
|
|
## Malwhisper-v1-small |
|
|
|
This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) fine-tuned on [IMASc dataset](https://www.kaggle.com/datasets/thennal/imasc). |
|
|
|
IMaSC is a Malayalam text and speech corpus made available by ICFOSS for the purpose of developing speech technology for Malayalam, particularly text-to-speech. The corpus contains 34,473 text-audio pairs of Malayalam sentences spoken by 8 speakers, totalling in approximately 50 hours of audio. |
|
|
|
The fine-tuned model on evaluating in the following dataset: |
|
|
|
**In Mozilla CommonVoice 11.0 dataset (Malayalam subset):** |
|
|
|
WER - 23.69 |
|
|
|
CER - 12.17 |
|
|
|
**In SMC Malayalam Speech Corpus dataset:** |
|
|
|
WER - 23.8 |
|
|
|
CER - 12.63 |
