import gradio as gr
from nemo.collections.asr.models import ASRModel

# Load the NeMo ASR model
model = ASRModel.from_pretrained("nvidia/canary-1b")
model.eval()

def transcribe(audio):
    if audio is None:
        raise gr.InterfaceError("Please provide some input audio")

    # Perform speech recognition
    transcription = model.transcribe([audio])

    return transcription[0]

audio_input = gr.inputs.Audio()

iface = gr.Interface(transcribe, audio_input, "text", title="ASR with NeMo Canary Model")
iface.launch()