Custom Workflow Automation
Emotion Tags
normalslowcryingsleepysighchuckle
Guided Emotion and Intonation
Industry Templates
Real EstateE-CommerceHealthcareBankingCustomer SupportCall CentersAppointment SchedulingEducationSalesRecruitment
Input Streaming for Lower Latency
Knowledgebase Integration
Launch Time
under 21 days
LLM-based Customizability
Model Tokenizer Type
Non-streaming (CNN-based) tokenizer
Natural Language Understanding
Open Source Release Planned
Orpheus Speech Models
Medium (3B)Small (1B)Tiny (400M)Nano (150M)
Penetration and Unit Testing
Personalized Interactions
Pretrained and Finetuned Models
Pretrained modelsFinetuned models
Real-time Appointment Booking
Sample Finetuning Scripts
Sliding Window Detokenizer
Streaming Inference Speed
Faster than playback on A100 40GB for 3B model
Strict Guardrails (No Hallucinations)
Training Data Volume
100k+ hours of speech, billions of text tokens