app.py

import gradio as gr

from transformers import pipeline

MODEL_NAME="ctaguchi/wav2vec2-large-xlsr-japlmthufielta-ipa1000-ns"
#MODEL_NAME="ginic/wav2vec-large-xlsr-en-ipa")

pipe = pipeline(task="automatic-speech-recognition", model=MODEL_NAME)

def predict(audio_in):
    return pipe(audio_in)["text"]


def launch_demo():
    with gr.Blocks() as demo:
        gr.Markdown(f"""
        # Automatic International Phonetic Alphabet Transcription
        This demo allows you to experiment with producing phonetic transcriptions of uploaded or recorded audio using the model '{MODEL_NAME}'.
        """)
        gr.Interface(fn=predict, inputs=gr.Audio(type="filepath"), outputs="text", allow_flagging="never")

    demo.launch()

if __name__ == "__main__":
    launch_demo()