Upload 3 files

app.py

@@ -39,24 +39,32 @@ model.load_state_dict(checkpoint)
 
 # %%
 def predict(input):
-    waveform, sr = librosa.load(input)
-    waveform = torch.from_numpy(waveform).unsqueeze(0)
-    waveform = torchaudio.transforms.Resample(sr, 16_000)(waveform)
-    inputs = feature_extractor(waveform, sampling_rate=feature_extractor.sampling_rate,
-                            max_length=16000, truncation=True)
-    tensor = torch.tensor(inputs['input_values'][0])
-    with torch.no_grad():
-        output = model(tensor)
-        logits = output['logits'][0]
-        label_id = torch.argmax(logits).item()
-    label_name = id2label[str(label_id)]
-
-    return label_name
+    if input == None:
+        return "Please input a valid file or record yourself by clicking the microphone"
+    elif input:
+        waveform, sr = librosa.load(input)
+        waveform = torch.from_numpy(waveform).unsqueeze(0)
+        waveform = torchaudio.transforms.Resample(sr, 16_000)(waveform)
+        inputs = feature_extractor(waveform, sampling_rate=feature_extractor.sampling_rate,
+                                max_length=16000, truncation=True)
+        tensor = torch.tensor(inputs['input_values'][0])
+        with torch.no_grad():
+            output = model(tensor)
+            logits = output['logits'][0]
+            label_id = torch.argmax(logits).item()
+        label_name = id2label[str(label_id)]
+
+        return label_name
+    else:
+        return "File is not valid"
 # %%
 demo = gr.Interface(
     fn=predict,
-    inputs=gr.Audio(source="microphone", type="filepath", label="Speak to classify your voice!"), # record audio, save in temp file to feed to inference func
-    outputs="text"
+    title="Audio Gender Classification",
+    description="Record your voice or upload an audio file to see what gender our model classifies it as",
+    inputs=gr.Audio(source="microphone", type="filepath", optional=False, label="Speak to classify your voice!"), # record audio, save in temp file to feed to inference func
+    outputs="text",
+    examples= [["male.mp3"], ["female.mp3"]]
 )
 
 # %%