import gradio as gr
from faster_whisper import WhisperModel

device = "cpu"
model_size = "base"
compute_type = "int8"

model = WhisperModel(model_size, device=device, compute_type=compute_type)
    
def transcribe(audio):
    segments, _ = model.transcribe(audio, beam_size=5)
    return "".join([segment.text for segment in segments])
 
gr.Interface(
    title = 'Fast Whisper for Speech Recognition', 
    description = 'This is a base version running on CPU with int8 compute type due to limited resources. These choices can slightly reduce accuracy.',
    fn=transcribe, 
    inputs=[
        gr.inputs.Audio(source="microphone", type="filepath")
    ],
    outputs=[
        "textbox"
    ]
).launch()