Qwen3 · Text To Speech
Design or pick a voice, type, and generate vivid speech.
Custom Voice · Voice Design · Multilingual
Synthesis Request
Custom Voice
Voice Design
Voice Clone
Text
Auto-split long text (≤500 chars per chunk)
Saved voice profile (skip ref audio/text)
None
Delete profile
Export .pt
Import .pt
Save current ref as profile
Save profile
Model (auto-picked by mode)
Language
Auto detect
Speaker (Custom Voice)
Pick a speaker
Style / Persona (optional for Custom Voice, required for Voice Design)
Save current designed voice as profile (.pt)
Save designed voice
Reference audio (URL / local path / base64)
Upload audio
🎙️ Start recording
Clear sample
Reference transcript (required unless x-vector only)
Use x-vector only (skip transcript; slightly lower quality)
Generate
Cancel
Output
Audio Export
Format
WAV
MP3
Download
Video Export
Style
Waveform
Spectrum
Pulse
Layout
Vertical 9:16
Square 1:1
Landscape 16:9
Include transcript
Export video
Download
Model Info
Model
-
Device
-