viVoice: Vietnamese Multi-Speaker Speech Synthesis with 100K+ Records

viVoice provides between 100,000 and 1,000,000 Vietnamese audio-text pairs for multi-speaker speech synthesis, released by capleaf in 2024. The dataset is specifically formatted for text-to-speech tasks and is distributed via Parquet files.

Use Cases

• Training multi-speaker text-to-speech (TTS) models using the Vietnamese audio recordings and text transcripts
• Voice cloning research across different speaker profiles
• Linguistic analysis of Vietnamese prosody and phonetics

Strengths

• Contains 100,000 to 1,000,000 audio-text pairs
• Supports multi-speaker synthesis for the Vietnamese language
• Available in Parquet format for efficient data handling

Limitations

• Restricted access requiring institutional or work email verification
• Non-commercial license (CC-BY-NC-SA-4.0) limits usage to research

Provenance

Source

capleaf

Freshness

Last updated July 2024

Geography

Vietnam