app.py

import torch
import gradio as gr
from transformers import pipeline

MODEL_NAME_V1 = "rngzhi/cs3264-project"
MODEL_NAME_V2 = "rngzhi/cs3264-project-v2"
BATCH_SIZE = 8
FILE_LIMIT_MB = 1000

device = 0 if torch.cuda.is_available() else "cpu"


def load_model(model_version):
    model_name = MODEL_NAME_V1 if model_version == 'Model-v1' else MODEL_NAME_V2
    return pipeline(
        task="automatic-speech-recognition",
        model=model_name,
        chunk_length_s=30,
        device=device,
    )

def transcribe(model_version, inputs, task):
    if inputs is None:
        raise gr.Error("No audio file submitted! Please upload or record an audio file before submitting your request.")

    pipe = load_model(model_version)
    text = pipe(inputs, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps=True)["text"]
    return text


demo = gr.Blocks()
mic_transcribe = gr.Interface(
    fn=transcribe,
    inputs=[gr.Dropdown(choices=['Model-v1', 'Model-v2'], label="Choose Model Version"), gr.Audio(sources="microphone", type="filepath")],
    outputs="text",
)

file_transcribe = gr.Interface(
    fn=transcribe,
    inputs=[gr.Dropdown(choices=['Model-v1', 'Model-v2'], label="Choose Model Version"), gr.Audio(sources="upload", type="filepath")],
    outputs="text",
    examples=[["Model-v2", "samples/sample1.WAV", "upload"], ["Model-v2", "samples/sample2.WAV", "upload"]]
)


with demo:
    gr.TabbedInterface([file_transcribe, mic_transcribe], ["Audio file", "Microphone"])
demo.launch(debug=True)