fix: slim deps, add packages.txt, drop runtime.txt, fix launch() binding

README.md

@@ -12,4 +12,4 @@ suggested_hardware: t4-small
 short_description: Singing-voice lyrics transcription via fine-tuned Whisper.
 ---
 
-See [GitHub](https://github.com/ram.duvvuri/autolyrics) for the full story.
+See [GitHub](https://github.com/ramduvvuri/autolyrics) for the full story.

app.py

@@ -158,4 +158,8 @@ with gr.Blocks(theme=THEME, css=CSS, title="AUTOLYRICS") as demo:
         outputs=[base_out, ft_out, base_meta, ft_meta, meta],
     )
 
-demo.queue(max_size=12).launch()
+demo.queue(max_size=12).launch(
+    server_name="0.0.0.0",   # Required: HF Spaces proxy expects this binding
+    server_port=7860,         # HF Spaces standard port
+    show_error=True,          # Surface tracebacks in the UI during debugging
+)

packages.txt

@@ -0,0 +1 @@
+ffmpeg

requirements.txt

@@ -1,19 +1,11 @@
-# ── Core ML stack ──────────────────────────────────────────────────────────
-# Remove +cpu suffix if the Space is always guaranteed a T4 GPU
-torch==2.3.1
-torchaudio==2.3.1
-transformers==4.47.1        # first release stable on Python 3.13; full Whisper support
+torch==2.7.1
+torchaudio==2.7.1
+transformers==4.47.1
 peft==0.13.2
-accelerate==0.34.2          # compatible with transformers 4.47.x
-
-# ── Gradio stack ───────────────────────────────────────────────────────────
 gradio==5.6.0
-pydantic==2.10.6            # CRITICAL: pins below pydantic 2.11 which breaks gradio_client 1.4.x
-                            # (gradio_client get_type: "if const in schema" crashes on bool)
-
-# ── HuggingFace ────────────────────────────────────────────────────────────
-huggingface_hub==0.26.3     # minimum required by peft 0.13.2 (0.25.2 is too old)
-
-# ── Audio ──────────────────────────────────────────────────────────────────
-soundfile==0.12.1
-sentencepiece==0.2.0
+gradio_client==1.4.3
+pydantic==2.10.6
+huggingface_hub==0.26.3
+accelerate==1.1.1
+sentencepiece==0.2.0
+soundfile==0.12.1

runtime.txt

@@ -1 +0,0 @@
-python-3.10