Upload app.py

app.py

@@ -0,0 +1,16 @@
+import nemo.collections.asr as nemo_asr
+import gradio as gr
+import time
+
+asr_model = nemo_asr.models.ASRModel.from_pretrained("nvidia/stt_en_conformer_transducer_xlarge")
+asr_model = asr_model.cuda()
+
+def transcribe(audio1, audio2):
+    mic_audio_transcription = asr_model.transcribe([audio1])[0][0] if audio1 else ""
+    file_audio_transcription = asr_model.transcribe([audio2])[0][0] if audio2 else ""
+    return mic_audio_transcription, file_audio_transcription
+
+gr.Interface(
+    fn=transcribe,
+    inputs=[gr.Audio(source="microphone", type="filepath"), gr.Audio(source="upload", type="filepath")],
+    outputs=["text", "text"]).launch()