Custom Deployment Options
Emotion Tags
normalslowcryingsleepysighchuckle
Enterprise Admin Controls
Guided Emotion and Intonation
Input Streaming for Lower Latency
Language/Country Support
30
LLM-based Customizability
Model Tokenizer Type
Non-streaming (CNN-based) tokenizer
Open Source Release Planned
Orpheus Speech Models
Medium (3B)Small (1B)Tiny (400M)Nano (150M)
Pretrained and Finetuned Models
Pretrained modelsFinetuned models
Real-Time AI Voice Translation
Sample Finetuning Scripts
Sliding Window Detokenizer
Streaming Inference Speed
Faster than playback on A100 40GB for 3B model
Supported Conferencing Apps
Any conferencing app
Training Data Volume
100k+ hours of speech, billions of text tokens
Ultra Low Latency
< 100ms