main.py

from flask import Flask, request, jsonify
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import numpy as np
import torch

app = Flask(__name__)
processor = Wav2Vec2Processor.from_pretrained("oyqiz/uzbek_stt")
model = Wav2Vec2ForCTC.from_pretrained("oyqiz/uzbek_stt")
SAMPLE_RATE = 16000

@app.route('/', methods=['GET'])
def index():
    return jsonify({"message": "Welcome to whisper uz!"})

@app.route('/transcribe', methods=['POST'])
def transcribe():
    data_frames = request.data
    audio_np = np.frombuffer(data_frames, dtype=np.int16)
    audio_np = audio_np / np.iinfo(np.int16).max
    inputs = processor(audio_np, sampling_rate=SAMPLE_RATE, return_tensors="pt")

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

    predicted_ids = torch.argmax(logits, dim=-1)
    transcription = processor.decode(predicted_ids[0])
    return transcription

if __name__ == '__main__':
    app.run(host='0.0.0.0', port=7860)