app.py

import gradio as gr
import torch
import torchaudio
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import librosa
import numpy as np
import re

processor = Wav2Vec2Processor.from_pretrained("facebook/wav2vec2-base-960h")
model = Wav2Vec2ForCTC.from_pretrained("facebook/wav2vec2-base-960h")
model.to("cpu")  #.to("cuda")

def transcribe(file_):
    arr_audio, _ = librosa.load(file_, sr=16000)
    input_values = processor(arr_audio, return_tensors="pt", padding="longest").input_values
    logits = model(input_values).logits  #model(input_values.to("cuda")).logits
    predicted_ids = torch.argmax(logits, dim=-1)
    transcription = processor.batch_decode(predicted_ids)
    return transcription[0].lower()


iface = gr.Interface(
    fn=transcribe, 
    inputs=gr.Audio(type="filepath"),
    outputs="text",
    title="Wave2Vec EN",
    description="Realtime demo for English speech recognition using a wave2vec model.",
)

iface.launch()