All your cloned and designed voices in one place. Play, manage, benchmark, and export them.
Voice library
Benchmarking voices...0 / 0
Elapsed -Avg -ETA -OK 0Slow 0Errors 0
1 Load or record source
Drop an audio / video file hereWAV · MP3 · OGG · FLAC · M4A · MP4 · MKV · WEBMor click to browse
Record a fresh sample
0:00
Input level-∞ dB
1.00x
Best peaks: -18 to -9 dB, never red.
Use this as the spoken script if you record.
Quiet room20 cm from micNo clippingNatural pace
If the microphone is blocked:
Chrome, Brave, Edge: click the lock/tune icon in the address bar, set Microphone to Allow, then reload.
Firefox: click the microphone or lock icon in the address bar, remove Blocked or choose Allow, then reload.
Safari: open Safari Settings, Websites, Microphone, then allow this site.
Requested device not found: choose or enable a microphone in your OS input settings, then reload.
Browsers require localhost or HTTPS for microphone access.
Load a file, paste a URL, or record a sample. Then trim, name, and save the voice.
Select 3-20 seconds for best cloning.
2 Name and save voice
Voices
Benchmark active voices?
About disabled voices: The Active toggle moves the complete voice package between
active_voices and hidden_voices. Qwen3-TTS should scan only active_voices.
To permanently remove a voice, delete its .wav and .reference.txt files directly.