Introduce modular transcription app: frontend + backend

README.md

@@ -11,7 +11,8 @@ tags:
   - audio
   - speech-to-text
   - whisper
-  - transcription
 models:
   - openai/whisper-small
+preload_from_hub:
+  - openai/whisper-small
 ---

app.py

@@ -1,43 +1,33 @@
+# app.py
 import gradio as gr
-import torch
-from transformers import pipeline
 
-# Load the Whisper model for speech recognition
-device = 0 if torch.cuda.is_available() else -1
-transcriber = pipeline(
-    "automatic-speech-recognition",
-    model="openai/whisper-small",
-    chunk_length_s=30,
-    device=device
-)
+# Import backend
+from backend import AudioTranscriber
+
+# Initialize the backend transcriber
+transcriber = AudioTranscriber()
 
 def transcribe_audio(audio_file):
-    """
-    Transcribe uploaded audio file.
-    Args:
-        audio_file (str): Path to the audio file
-    Returns:
-        str: Transcribed text
-    """
-    if audio_file is None:
-        return "Please upload an audio file."
-    
-    # Run transcription
-    result = transcriber(audio_file)
-    return result["text"]
+    return transcriber.transcribe(audio_file)
 
-# Define the Gradio interface
+# Define Gradio Interface
 demo = gr.Interface(
     fn=transcribe_audio,
-    inputs=gr.Audio(sources=["upload", "microphone"], type="filepath"),
-    outputs=gr.Textbox(label="Transcription", lines=8),
+    inputs=gr.Audio(
+        label="Upload or Record Audio",
+        sources=["upload", "microphone"],
+        type="filepath"
+    ),
+    outputs=gr.Textbox(label="Transcription", lines=6, placeholder="Transcription will appear here..."),
     title="🎙️ Audio Transcription with Whisper",
-    description="Upload an audio file or record directly to transcribe it using OpenAI's Whisper model.",
+    description="Upload an audio file or record your voice. The backend will transcribe it using Whisper.",
     examples=[
-        ["example.wav"]  # Optional: add a sample audio file named example.wav
+        ["example.wav"]  # Optional: include a sample file
     ],
+    cache_examples=False,
     allow_flagging="never"
 )
 
 # Launch the app
-demo.launch()
+if __name__ == "__main__":
+    demo.launch()

backend.py

@@ -0,0 +1,31 @@
+# backend.py
+from transformers import pipeline
+import torch
+
+class AudioTranscriber:
+    def __init__(self, model_name="openai/whisper-small"):
+        self.device = 0 if torch.cuda.is_available() else -1
+        print(f"Loading model '{model_name}' on device: {'GPU' if self.device == 0 else 'CPU'}")
+        self.pipe = pipeline(
+            "automatic-speech-recognition",
+            model=model_name,
+            chunk_length_s=30,
+            device=self.device
+        )
+
+    def transcribe(self, audio_path):
+        """
+        Transcribe an audio file.
+        Args:
+            audio_path (str): Path to the audio file
+        Returns:
+            str: Transcribed text
+        """
+        if audio_path is None:
+            return "No audio file provided."
+
+        try:
+            result = self.pipe(audio_path)
+            return result.get("text", "").strip()
+        except Exception as e:
+            return f"Transcription error: {str(e)}"

requirements.txt

@@ -0,0 +1,3 @@
+torch==2.3.1
+transformers==4.41.2
+gradio==4.25.0