chore: upgrade Gradio to 6.2.0 and align pyannote.audio dependencies

- Upgrade gradio from >=5.49.1 to >=6.2.0 in requirements.txt and pyproject.toml
- Align pyannote.audio from ==3.1.1 to >=4.0.0 in pyproject.toml to match requirements.txt
- Resolve compatibility issue with torchaudio 2.8+ where set_audio_backend was removed
- pyannote.audio 4.0+ is compatible with the modern torchaudio API

Dependencies updated:
- gradio: >=5.49.1 → >=6.2.0 (both files)
- pyannote.audio: ==3.1.1 → >=4.0.0 (pyproject.toml only)

Note: Gradio 6.2.0 has a known Safari UI rendering issue with certain file sizes
(blob URL + range requests), but upload/processing functionality works correctly.

pyproject.toml

@@ -23,12 +23,12 @@ dependencies = [
     "torch>=2.4.0",
     "torchaudio>=2.4.0",
     "transformers>=4.35.0",
-    "gradio>=5.49.1",
+    "gradio>=6.2.0",
     "spaces>=0.28.3",  # HuggingFace ZeroGPU support
 
     # HuggingFace models
     "huggingface-hub>=0.16.0",
-    "pyannote.audio==3.1.1",
+    "pyannote.audio>=4.0.0",
 
     # Audio processing
     "librosa>=0.10.0",

requirements.txt

@@ -5,7 +5,7 @@ spaces>=0.28.3
 torch>=2.4.0
 torchaudio>=2.4.0
 transformers>=4.35.0
-gradio>=5.49.1
+gradio>=6.2.0
 
 # HuggingFace models
 huggingface-hub>=0.16.0