Upload 5 files

app.py

@@ -0,0 +1,13 @@
+import gradio as gr
+from inference import predict
+
+def predict_from_audio(audio):
+    return predict(audio)["language"]
+
+demo = gr.Interface(
+    fn=predict_from_audio,
+    inputs=gr.Audio(type="filepath"),
+    outputs="label",
+    title="Custom Language Identifier",
+    description="Upload audio → detects English, Mandarin, Tamil, Malay, Iban, Bukar Sadong"
+)

inference.py

@@ -0,0 +1,17 @@
+import torch, joblib, json
+from speechbrain.pretrained import EncoderClassifier
+from your_module import Config, AudioProcessor, LanguageIdentifier
+
+# Load classifier + config
+custom_classifier = joblib.load("polar_lid_classifier.pkl")
+label_map = json.load(open("label_map.json"))
+
+config = Config()
+config.label_map = label_map
+
+identifier = LanguageIdentifier(config)
+identifier.load_vox_model()
+identifier.custom_classifier = custom_classifier
+
+def predict(audio_path: str):
+    return identifier.predict(audio_path)

label_map.json

@@ -0,0 +1 @@
+{"iban": 0, "bukar_sadong": 1, "malay": 2}

polar_lid_classifier.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7237502bf2f5a7d5cdde1b450445ace9a3e20230bf2358053ea83bb2325fe7e
+size 7055

requirements.txt

@@ -0,0 +1,5 @@
+torch
+speechbrain
+scikit-learn
+joblib
+gradio