Initial commit for Gradio API

app.py

@@ -0,0 +1,68 @@
+import os
+import shutil
+import torchaudio
+import gradio as gr
+from speechbrain.inference import SpeakerRecognition
+from fastapi import HTTPException
+
+# Initialize the speaker verification model
+speaker_verification = SpeakerRecognition.from_hparams(
+    source="speechbrain/spkrec-ecapa-voxceleb",
+    savedir="tmp_model"
+)
+
+# Temporary folder to save uploaded files
+UPLOAD_FOLDER = "uploaded_audio"
+os.makedirs(UPLOAD_FOLDER, exist_ok=True)
+
+# Function to calculate similarity score
+def get_similarity(audio_path1: str, audio_path2: str):
+    try:
+        # Load audio files
+        signal1, _ = torchaudio.load(audio_path1)
+        signal2, _ = torchaudio.load(audio_path2)
+
+        # Get similarity score and prediction
+        score, prediction = speaker_verification.verify_batch(signal1, signal2)
+        return float(score), "Yes" if prediction else "No"
+    except Exception as e:
+        return str(e), None
+    finally:
+        # Clean up temporary files
+        if os.path.exists(audio_path1):
+            os.remove(audio_path1)
+        if os.path.exists(audio_path2):
+            os.remove(audio_path2)
+
+# API function to compare voices
+def compare_voices(file1, file2):
+    # Save uploaded files temporarily
+    file1_path = os.path.join(UPLOAD_FOLDER, file1.name)
+    file2_path = os.path.join(UPLOAD_FOLDER, file2.name)
+
+    with open(file1_path, "wb") as f1:
+        f1.write(file1.read())
+    with open(file2_path, "wb") as f2:
+        f2.write(file2.read())
+
+    # Get similarity score
+    score, is_same_user = get_similarity(file1_path, file2_path)
+
+    if is_same_user is None:
+        return "Error: " + score  # This will return the error message
+
+    return {"Similarity Score": f"{score:.4f}", "Same User Prediction": is_same_user}
+
+# Create Gradio Interface for the API
+api = gr.Interface(
+    fn=compare_voices,
+    inputs=[
+        gr.Audio(source="upload", type="file", label="First Audio File"),
+        gr.Audio(source="upload", type="file", label="Second Audio File")
+    ],
+    outputs="json",  # Output results as JSON
+    live=False  # No live interface, just the API
+)
+
+# Launch the API as an HTTP server
+api.queue().launch(server_name="0.0.0.0", server_port=8080, share=False)

requirements.txt

@@ -0,0 +1,9 @@
+fastapi
+uvicorn
+numpy
+scikit-learn
+joblib
+torchaudio
+speechbrain
+python-multipart
+gradio