vocca_ai/ai_response.py

from transformers import AutoModelForSeq2SeqLM, AutoTokenizer


model_name = "google/flan-t5-large"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)

def generate_call_summary(transcript):
    """
    Generates a structured and useful summary of the call.
    """
    input_text = f"Summarize this medical call conversation:\n{transcript}"
    inputs = tokenizer(input_text, return_tensors="pt", truncation=True)
    outputs = model.generate(**inputs, max_length=100, min_length=20, length_penalty=2.0, num_beams=5)

    return tokenizer.decode(outputs[0], skip_special_tokens=True)


if __name__ == "__main__":
    sample_text = "Patient: Hi, I need to schedule an appointment as soon as possible. I’ve been feeling really weak and dizzy for the past few days."
    print(f"Call Summary: {generate_call_summary(sample_text)}")