Repository Issue Activity (beta)

#80 Base model voice clone sometimes change speaker

#350 Multi-concurrency inference not supported

#347 Verify evals on Papers with Code

#365 finetuning/sft_12hz.py: checkpoint save is non-atomic (can silently leave base weights) and peaks at 2x model size in host RAM

#239 Inconsistent speaking rate in long text generation with Qwen3-TTS-12Hz-1.7B-Base

#273 多音字问题！！！

#333 Nan logits with flash attention backend

#323 Fine-tuning Qwen3-TTS-12Hz-1.7B-Base for Bengali: Foreign/Hindi Accent Bias and Lack of Naturalness

#340 AMD MI300X - Qwen/Qwen3-TTS-12Hz-1.7B-Base - Voice Cloning Segmentation fault (core dumped)

#328 使用音色克隆模型时，无法读准时间字符串

#314 [Performance Regression] Lower speaker similarity (SIM) after SFT than zero-shot on Qwen3-TTS-1.7B

#341 Voice-clone (ICL) generation echoes the tail of the reference audio before the requested text

#359 Feature request: native Italian preset voices (male + female)

#179 Finetuning Base results in progressively faster speech with every epoch

#318 [Bug] generate_voice_clone() hangs indefinitely on mixed-script Thai inputs (Qwen3-TTS-12Hz-1.7B-Base)

#337 The talker_hidden_states not obtained correctly

#55 The audio generated for Japanese and Korean is incomplete.

#290 Speaking Rate Ignored in Clone. It is Always Faster than Cloned Voice and Time Compressed

#335 An error occurred while fine-tuning the 0.6B model using the Fine-Tuning Script within the repository

#258 Massive distortion within audio file for seemingly no reason

#39 Fine-tuned model outputs noise despite decreasing training loss

#141 How to generate Cantonese (粤语)?

#34 Release Qwen3-TTS 25Hz models and tokenizer on Hugging Face

#89 Very slow inference on 5090

#298 在每次合成的时候，输入不变如何能得到相同的输出//使用Qwen3-TTS-12Hz-1.7B-CustomVoice/）