Upload an audio file or record from your microphone to transcribe it using the ABR-AI model with custom inference.