faster-qwen3-tts

CUDA GRAPHS
Clone — match a voice from a reference clipLOAD ▶
Simple
Advanced
Advanced cloning uses the transcript of the reference clip for higher accuracy. Upload or record to auto-transcribe.
Custom — use a built-in speakerLOAD ▶
Voice Design — describe any voiceLOAD ▶
TTFA
Client
RTF
Dur
Buf
Settings
Model
Generation
Streaming
Non-streaming
8
Sampling
0.9
50
1.05
Recording