Conversational Speech Generation
Dataset Size
1 million hours
In-Person Conversation Translation
Language and Accent Variants
Model Sizes
Tiny: 1B backbone, 100M decoderSmall: 3B backbone, 250M decoderMedium: 8B backbone, 300M decoder
Multiple Speaker Handling
No App Needed for Invitees
No Need for Human Interpreter
Objective Metrics
Word Error RateSpeaker SimilarityHomograph DisambiguationPronunciation Consistency
Partial Multilingual Support Planned
Planned for 20+ languages
Real-Time Translation Speed
Slight delay
Speech to Speech Translation
Subjective Metrics
Comparative Mean Opinion Score
Supported Languages
عربيDeutschEnglishEspañolFrançaisIndonesia日本語한국인РусскийแบบไทยTiếng ViệtPortuguêsPolskiУкраїнська简体中文繁体中文
Supports Daily Life, Travel, and Business
ImmigrantsTravelersBusiness
Supports Multiple Languages
150
Total Supported Languages
150