Qwen Alibaba Cloud: Qwen3-TTS-12Hz-1.7B-CustomVoice

Qwen's 1.7B open-weight text-to-speech model with a 12 Hz acoustic token rate. Synthesizes 24 kHz 16-bit mono speech from a fixed set of built-in speaker voices in ten languages, with a speaking-rate multiplier and optional natural-language style guidance. The same checkpoint also backs reference-audio voice cloning, which is a separate hosted surface and is not part of this card.