Upload 2 files

app.py

@@ -0,0 +1,41 @@
+import modal
+
+app = modal.App("whisper-app")
+
+# Define container with required packages
+image = (
+    modal.Image.debian_slim()
+    .pip_install("openai-whisper", "requests")
+    .apt_install("ffmpeg")
+)
+
+@app.cls(image=image)
+class WhisperModel:
+    def __init__(self):
+        self.model = None
+
+    @modal.enter()  # ✅ This is critical: use @modal.enter() not just __enter__
+    def init(self):
+        import whisper
+        self.model = whisper.load_model("base")
+
+    @modal.method()
+    def transcribe(self, audio_url):
+        import requests
+
+        # Download audio file
+        response = requests.get(audio_url)
+        with open("audio.wav", "wb") as f:
+            f.write(response.content)
+
+        # Run transcription
+        result = self.model.transcribe("audio.wav")
+        return result["text"]
+
+@app.local_entrypoint()
+def main():
+    url = "https://huggingface.co/datasets/Narsil/asr_dummy/resolve/main/mlk.flac"
+
+    model = WhisperModel()
+    result = model.transcribe.remote(url)
+    print(result)

gradio_app.py

@@ -0,0 +1,13 @@
+import gradio as gr
+import modal
+
+# Load your deployed Modal class
+WhisperModel = modal.Cls.from_name("whisper-app", "WhisperModel")
+
+# Define function to call remote transcribe
+def transcribe_audio(url):
+    instance = WhisperModel()  # creates an instance
+    return instance.transcribe.remote(url)
+
+# Launch Gradio UI
+gr.Interface(fn=transcribe_audio, inputs="text", outputs="text").launch()