app.py

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

# Load model and processor once
processor = Wav2Vec2Processor.from_pretrained("facebook/wav2vec2-large-960h-lv60-self")
model = Wav2Vec2ForCTC.from_pretrained("facebook/wav2vec2-large-960h-lv60-self")

def transcribe(audio):
    # audio is a tuple: (sample_rate, numpy_array)
    sample_rate, waveform = audio
    # Convert waveform to float32 and process
    input_values = processor(waveform, sampling_rate=sample_rate, return_tensors="pt", padding="longest").input_values
    with torch.no_grad():
        logits = model(input_values).logits
    predicted_ids = torch.argmax(logits, dim=-1)
    transcription = processor.batch_decode(predicted_ids)
    return transcription[0]

# Build Gradio interface
iface = gr.Interface(
    fn=transcribe,
    inputs=gr.Audio(source="microphone", type="numpy"),
    outputs="text",
    title="Wav2Vec2 ASR",
    description="Record or upload audio, and get transcription using Wav2Vec2 large model."
)

iface.launch()