--- language: - ko license: cc-by-4.0 task_categories: - text-to-speech tags: - tts - korean - speech - gemini size_categories: - n<1K dataset_info: features: - name: audio dtype: audio: sampling_rate: 24000 - name: id dtype: int64 - name: category dtype: string - name: sub dtype: string - name: text dtype: string - name: difficulty dtype: string - name: instruct dtype: string splits: - name: train num_bytes: 30261995 num_examples: 120 download_size: 30256510 dataset_size: 30261995 configs: - config_name: default data_files: - split: train path: data/train-* --- # Korean TTS Training Dataset 한국어 TTS 학습용 음성 데이터셋 (120문장) ## Dataset Description Google Gemini `gemini-2.5-pro-preview-tts` 모델 (Voice: Zephyr)로 생성한 한국어 음성 데이터셋입니다. 발음, 운율, 감정, 억양 등 다양한 카테고리를 포함합니다. ### Audio Format - Format: WAV (PCM) - Sample Rate: 24,000 Hz - Bit Depth: 16-bit - Channels: Mono ## Categories ### 발음/운율 (70문장) | Category | Count | Description | |----------|-------|-------------| | 1-숫자 | 10 | 숫자, 전화번호, 날짜, 금액 | | 2-겹받침 | 10 | 겹받침, ㅎ받침 발음 | | 3-음운변동 | 10 | 연음, 경음화, 비음화, 유음화 | | 4-모음ㄹ | 10 | 이중모음, ㄹ발음, 두음법칙 | | 5-외래어 | 10 | 외래어, 영어 혼용 | | 6-긴문장 | 10 | 긴 문장, 호흡 | | 8-특수 | 10 | 구개음화, 의성어, 의태어 | ### 감정/억양 (50문장) | Category | Count | Description | |----------|-------|-------------| | 7A-억양 | 13 | 의문문, 명령, 나열, 강조 등 | | 7B-긍정감정 | 12 | 기쁨, 감동, 감사, 안도, 설렘 | | 7C-부정감정 | 13 | 슬픔, 분노, 짜증, 불안, 실망 | | 7D-복합전환 | 12 | 감정 변화, 반어, 망설임 | ## Usage ```python from datasets import load_dataset ds = load_dataset("daje/korean-tts-training") ``` ## Fields | Field | Description | |-------|-------------| | `file_name` | 오디오 파일명 | | `id` | 문장 고유 ID | | `category` | 카테고리 | | `sub` | 세부 카테고리 | | `text` | 한국어 문장 | | `difficulty` | 발음/억양 난이도 설명 | | `instruct` | TTS 생성 시 사용한 영어 프롬프트 |