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("cuda")

def transcribe(file_):
    arr_audio, _ = librosa.load(file_, sr=16000)
    inputs = processor(arr_audio, sampling_rate=16_000, return_tensors="pt", padding=True)

    with torch.no_grad():
        logits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits

    pred_ids = torch.argmax(logits, dim=-1)
    text = processor.batch_decode(pred_ids)[0]
    return text.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()