Update app.py

app.py

@@ -1,3 +1,25 @@
-import gradio as gr
-
-gr.Interface.load("models/nguyenanh2803/stt_conformer_large_na").launch()
+import gradio as gr
+import nemo.collections.asr as nemo_asr
+
+# Load the pretrained model
+conformer_large = nemo_asr.models.ASRModel.from_pretrained("nguyenanh2803/stt_conformer_large_na")
+
+# Define the transcription function
+def transcribe_audio(audio):
+    # audio is a tuple of (sample_rate, audio_data)
+    sample_rate, audio_data = audio
+    # Transcribe the audio file
+    transcription = conformer_large.transcribe([audio_data])
+    return transcription[0]
+
+# Create the Gradio interface
+interface = gr.Interface(
+    fn=transcribe_audio, 
+    inputs=gr.inputs.Audio(source="microphone", type="numpy", label="Speak or Upload an Audio File"), 
+    outputs="text",
+    title="Conformer Large ASR",
+    description="Transcribe audio using NVIDIA NeMo's Conformer Large model."
+)
+
+# Launch the interface
+interface.launch()