import gradio as gr
from transformers import pipeline
import numpy as np
import librosa

transcriber = pipeline("automatic-speech-recognition", model="Oysiyl/w2v-bert-2.0-ukrainian-colab-CV16.0")

def transcribe(audio):
    sr, y = audio
    if sr != 16000:
        y = librosa.resample(y, orig_sr=sr, target_sr=16000)
    y = y.astype(np.float32)
    y /= np.max(np.abs(y))

    return transcriber({"sampling_rate": sr, "raw": y})["text"]


demo = gr.Interface(
    transcribe,
    gr.Audio(sources=["upload", "microphone"]),
    outputs="text",
    title="Automatic Speech Recognition for Ukrainian language demo",
    description="Click on the example below, upload audio from file or say something in microphone!",
    examples=[["examples/asr_example.wav"], ["examples/tts_example.wav"]],
    cache_examples=True
)

demo.launch()