Create app.py

app.py

@@ -0,0 +1,53 @@
+import gradio as gr
+from transformers import pipeline
+import torch
+
+# Initialize ASR and classifier pipelines
+model_asr = "kairaamilanii/whisper-mind14-enUS"
+model_class = "kairaamilanii/RoBERTa-minds-14-classifier-intent"
+
+transcriber = pipeline(
+    "automatic-speech-recognition",
+    model=model_asr,
+    chunk_length_s=30,
+    device="cuda:0" if torch.cuda.is_available() else "cpu"
+)
+classifier = pipeline("text-classification", model=model_class)
+
+intent_classes = {
+    0: 'abroad',
+    1: 'address',
+    2: 'app_error',
+    3: 'atm_limit',
+    4: 'balance',
+    5: 'business_loan',
+    6: 'card_issues',
+    7: 'cash_deposit',
+    8: 'direct_debit',
+    9: 'freeze',
+    10: 'latest_transactions',
+    11: 'joint_account',
+    12: 'high_value_payment',
+    13: 'pay_bill'
+}
+
+# Function to process audio
+def process_audio(audio):
+    # Transcribe the audio
+    text_asr = transcriber(audio)['text']
+    # Classify the intent
+    intent_class = classifier(text_asr)
+    label_index = int(intent_class[0]['label'].split('_')[1])
+    intent_name = intent_classes.get(label_index, "Unknown")
+    return text_asr, intent_name
+
+# Create Gradio interface
+iface = gr.Interface(
+    fn=process_audio,
+    inputs=gr.Audio(source="upload", type="filepath"),
+    outputs=["text", "text"],
+    title="ASR and Intent Classification",
+    description="Upload an audio file to get transcription and intent classification."
+)
+
+iface.launch()