Downloads · 30 days
23
31% of all-time downloads
seastar105/whisper-base-kspon
whisper-base-kspon is a automatic speech recognition model from seastar105. Use it when you need speech turned into text. It is set up for transformers.
OpenAI의 whisper-base 모델을 아래 데이터셋으로 학습한 모델입니다. - 한국어 음성 (https://www.aihub.or.kr/aihubdata/data/view.do?currMenu=115&topMenu=100&aihubDataSe=realm&dataSetSn=123)
Downloads · 30 days
23
31% of all-time downloads
All-time downloads
74
Public
Parameters
72.6M
1.2 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors290 MB · 98%
From the Hugging Face model README
OpenAI의 whisper-base 모델을 아래 데이터셋으로 학습한 모델입니다.
train_steps: 5000
warmup_steps: 500
lr scheduler: linear warmup cosine decay
max learning rate: 1e-4
batch size: 128
max_grad_norm: 1.0
adamw_beta1: 0.9
adamw_beta2: 0.98
https://github.com/rtzr/Awesome-Korean-Speech-Recognition
위 레포지토리에서 주요 영역별 회의 음성을 제외한 테스트셋 결과입니다. 아래 테이블에서 whisper_base_kspon이 본 모델 성능입니다.
+--------------------------+----------+-----------+---------------+------------+---------------+--------------+-------------+-------------+
| Model | cv_15_ko | fleurs_ko | kcall_testset | kconf_test | kcounsel_test | klec_testset | kspon_clean | kspon_other |
+--------------------------+----------+-----------+---------------+------------+---------------+--------------+-------------+-------------+
| whisper_base | 21.16 | 11.89 | 42.56 | 27.62 | 22.24 | 28.65 | 30.41 | 27.02 |
| whisper_base_kspon | 26.63 | 13.95 | 42.05 | 29.61 | 26.21 | 28.72 | 12.58 | 13.48 |
| whisper_base_komix | 15.42 | 7.16 | 20.86 | 14.24 | 12.64 | 13.44 | 12.26 | 12.12 |
| whisper_turbo | 5.38 | 3.95 | 5.89 | 9.77 | 4.21 | 9.27 | 16.49 | 13.54 |
+--------------------------+----------+-----------+---------------+------------+---------------+--------------+-------------+-------------+