datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
QWEN3-TTS-Voice-Design-100-Japanese-Female-Designed-Voices100 Japanese Female Designed Voices by Qwen3-TTS-12Hz-1.7B-VoiceDesign
Note: Contains frequent misreadings. Correct reading data is not provided.
AI Generation: The 100 styles were automatically generated by AI, so there may be some overlaps or duplicates.
voice is designed by Japanese Prompt(see styles_jp.txt)
Dataset: 300 audio clips (100 styles × 3 iterations).
Structure: design1–design3 represent each iteration. Each output is unique.
Fixes: Replaced one instance of a "complete error"… See the full description on the dataset page: https://huggingface.co/datasets/Akjava/QWEN3-TTS-Voice-Design-100-Japanese-Female-Designed-Voices.QWEN3-TTS-Voice-Design-100-Japanese-Female-Designed-Voices100 Japanese Female Designed Voices by Qwen3-TTS-12Hz-1.7B-VoiceDesign
Note: Contains frequent misreadings. Correct reading data is not provided.
AI Generation: The 100 styles were automatically generated by AI, so there may be some overlaps or duplicates.
voice is designed by Japanese Prompt(see styles_jp.txt)
Dataset: 300 audio clips (100 styles × 3 iterations).
Structure: design1–design3 represent each iteration. Each output is unique.
Fixes: Replaced one instance of a "complete error"… See the full description on the dataset page: https://huggingface.co/datasets/196vm3/QWEN3-TTS-Voice-Design-100-Japanese-Female-Designed-Voices.voice-design-bench-50voice-design-bench-50-dnsmos
Model
Mean
Min
Max
Qwen
3.262
2.480
3.619
Echo
3.217
2.158
3.588
Omni
3.190
0.988
3.629
VoiceDesign3
VoiceDesign
Vietnamese TTS dataset synthesized
voice_designA dataset for voice design.
~30K samples of voice design data (LAION)