from flask import Flask, request, jsonify
from transformers import AutoProcessor, AutoModelForSpeechSeq2Seq
from transformers import pipeline
import numpy as np
import json
app = Flask(__name__)
model = AutoModelForSpeechSeq2Seq.from_pretrained("GitNazarov/whisper-small-pt-3-uz")
processor = AutoProcessor.from_pretrained("GitNazarov/whisper-small-pt-3-uz")

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

@app.route('/transcribe', methods=['POST'])
def transcribe():
    data = request.json['data']
    data = json.loads(data)
    tensor_data = np.array(data)
    inputs = processor(tensor_data, return_tensors="pt", sampling_rate=16000, max_new_tokens=100)
    input_features = inputs.input_features
    generated_ids = model.generate(inputs=input_features)

    transcription = processor.batch_decode(generated_ids, skip_special_tokens=True)
    transcription = ''.join(transcription)
    return str(transcription), {'Content-Type': 'application/json'}

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