Transcribe, ChatGPT, gTTS

app.py

from flask import Flask, request, jsonify, send_file
import tempfile
import whisper
import gtts
import os
from openai import OpenAI
client = OpenAI()

conversation = []

system_prompt = """
You are a helpful assistant to help people learn Spanish. 
Your task is to role play a native conversation with the user. 
The transcription mightn't be perfect, but you should be able to understand the gist of what they're saying. 
You're very concise and complementary. After responding, you should keep the conversation going by asking a question."""

app = Flask(__name__)
model = whisper.load_model("base")

@app.route('/chat', methods=['POST'])
def chat():
    if 'file' not in request.files:
        return "No file part", 400

    file = request.files['file']
    if file.filename == '':
        return "No selected file", 400

    if file:
        temp_path = save_temp_audio_file(file)
        audio_transcription = transcribe(temp_path)
        conversation.append(f'User: {audio_transcription}')
        response = respond(0, system_prompt)
        conversation.append(f'Assistant: {response}')
        response_audio = say(response)
        return send_file(response_audio, as_attachment=True)

def save_temp_audio_file(audio_file):
    temp_file, temp_path = tempfile.mkstemp(suffix=".wav")
    os.close(temp_file)
    audio_file.save(temp_path)
    return temp_path

def transcribe(audio_file):
    result = model.transcribe(audio_file, language='es')
    return result

def respond(temperature, system_prompt):
    response = client.chat.completions.create(
        model="gpt-4",
        temperature=temperature,
        messages=[
            {
                "role": "system",
                "content": system_prompt
            },
            {
                "role": "user",
                "content": 'Conversation so far:' + "\n ".join(conversation) + 'Assistant: '
            }
        ]
    )
    return response.choices[0].message.content

def say(text):
    tts = gtts.gTTS(text, lang='es')
    filename = "response.mp3"
    tts.save(filename)
    return filename

if __name__ == '__main__':
    app.run(debug=True)