app.py

import os
import gradio as gr
from transformers import pipeline

pipeline = pipeline(task="automatic-speech-recognition", model="jonatasgrosman/wav2vec2-large-xlsr-53-german")
#pipeline = pipeline(task="automatic-speech-recognition", model="openai/whisper-large")

def transcribeFile(audio_path : str) -> str:
    transcription = pipeline(audio_path)
    return transcription["text"]

def transcribeMic(audio):
    sr, data = audio
    transcription = pipeline(data)
    return transcription["text"]

app1 = gr.Interface(
    fn=transcribeFile,
    inputs=gr.inputs.Audio(label="Upload audio file", type="filepath"),
    outputs="text"
)


app2 = gr.Interface(
    fn=transcribeMic,
    inputs="microphone",
    outputs="text"
)

demo = gr.TabbedInterface([app1, app2], ["Audio File", "Microphone"])

if __name__ == "__main__":
    demo.launch()