yt-data-1/synthetic-data-20k

--- dataset_info: features: - name: audio dtype: audio - name: text dtype: string - name: speaker_id dtype: string splits: - name: train num_examples: 20000 configs: - config_name: default data_files: - split: train path: data/train-*.parquet --- # Synthetic Data JA 20K synthetic Japanese conversational TTS audio samples. - **Speakers**: 20,000 unique speakers - 10,000 VoiceDesign speakers (speaker_00001 – speaker_10000) - 10,000 voice-cloned speakers (spk_clone_00001 – spk_clone_10000) - **Total samples**: 20,000 - **Sample rate**: 48 kHz - **Format**: WAV embedded in Parquet - **Language**: Japanese