Trending storyDeveloping

Armin Ronacher details custom voice cloning with Qwen3-TTS

1 article1 sourcesince Oct 30Last article 1h ago · Oct 30, 9:05 PM ET

Overview

AISummary of 1 article

Armin Ronacher says Qwen3-TTS Base can clone a voice from about 28 seconds of audio and extract a roughly 2048-number speaker embedding.

According to Ronacher, feeding that embedding into the CustomVoice model lets the cloned voice follow instructions such as sad or whispering, though he reports the results are less stable than with the built-in speakers. He also describes mixing embeddings and shifting perceived gender by moving along an average female-minus-male direction. The report comes from a post shared on X by meng shao.

Written by AI from the articles below · updated Oct 30, 9:25 PM ET

Check the sources:

Latest developmentOct 30, 9:05 PM ET

Armin Ronacher explains building custom voices with Qwen3-TTS

Article timeline

The articles in this story. Times are ET.

Oct 30

9:05 PM
meng shao@shao__mengX
Armin Ronacher explains building custom voices with Qwen3-TTS
AIArmin Ronacher says Qwen3-TTS Base can clone a voice from about 28 seconds of audio and extract a roughly 2048-number speaker embedding. He reports that feeding that embedding into CustomVoice lets the cloned voice take instructions such as sad or whispering, though results are less stable than with built-in speakers. He also describes mixing embeddings and shifting gender perception by moving along an average female-minus-male direction.

Heat Trend

Not enough continuous observations to show a trend yet.