import gradio as gr
from transformers import Wav2Vec2Processor, Wav2Vec2ForCTC
import torch
import phonemizer
import librosa

processor = Wav2Vec2Processor.from_pretrained("facebook/wav2vec2-xlsr-53-espeak-cv-ft")
model = Wav2Vec2ForCTC.from_pretrained("facebook/wav2vec2-xlsr-53-espeak-cv-ft")

waveform, sample_rate = librosa.load('harvard.wav', sr=16000) # Downsample 44.1kHz to 8kHz

input_values = processor(waveform, sampling_rate=sample_rate, return_tensors="pt").input_values

with torch.no_grad():
   logits = model(input_values).logits

predicted_ids = torch.argmax(logits, dim=-1)
transcription = processor.batch_decode(predicted_ids)

def showTranscription(transcription):
   return transcription

iface = gr.Interface(fn=showTranscription, inputs="text", outputs="text")
iface.launch()