Upload folder using huggingface_hub

app.py

@@ -184,10 +184,12 @@ def build_ui():
         gr.Markdown(
             """
 # Qwen3-TTS Demo
+
 A unified Text-to-Speech demo featuring three powerful modes:
 - **Voice Design**: Create custom voices using natural language descriptions
 - **Voice Clone (Base)**: Clone any voice from a reference audio
 - **TTS (CustomVoice)**: Generate speech with predefined speakers and optional style instructions
+
 Built with [Qwen3-TTS](https://github.com/QwenLM/Qwen3-TTS) by Alibaba Qwen Team.
 """
         )
@@ -329,6 +331,7 @@ Built with [Qwen3-TTS](https://github.com/QwenLM/Qwen3-TTS) by Alibaba Qwen Team
         gr.Markdown(
             """
 ---
+
 **Note**: This demo uses HuggingFace Spaces Zero GPU. Each generation has a time limit.
 For longer texts, please split them into smaller segments.
 """
@@ -339,4 +342,4 @@ For longer texts, please split them into smaller segments.
 
 if __name__ == "__main__":
     demo = build_ui()
-    demo.launch()
+    demo.launch()

qwen_tts/cli/demo.py

@@ -630,4 +630,4 @@ def main(argv=None) -> int:
 
 
 if __name__ == "__main__":
-    raise SystemExit(main())
+    raise SystemExit(main())

qwen_tts/core/models/modeling_qwen3_tts.py

@@ -2243,4 +2243,4 @@ __all__ = [
     "Qwen3TTSTalkerForConditionalGeneration",
     "Qwen3TTSPreTrainedModel",
     "Qwen3TTSTalkerModel",
-]
+]

qwen_tts/inference/qwen3_tts_model.py

@@ -871,4 +871,4 @@ class Qwen3TTSModel:
         supported = self._supported_languages_set()
         if supported is None:
             return None
-        return sorted(supported)
+        return sorted(supported)

requirements.txt

@@ -10,5 +10,4 @@ sox
 onnxruntime
 spaces
 torch
-numpy
-kernels
+numpy