py/api.py

@@ -4,6 +4,7 @@ from transformers import AutoTokenizer, LlamaForCausalLM
 
 
 class API:
+    # This method processes a message via transformers. (NOT FINISHED!)
     @staticmethod
     def process_text_transformers(prompt, model):
         model = LlamaForCausalLM.from_pretrained(model)
@@ -14,6 +15,7 @@ class API:
         generate_ids = model.generate(inputs.input_ids, max_length=30)
         return tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
 
+    # This method processes a message via ollama
     @staticmethod
     def process_text_local(prompt, model):
         ollama_url = "http://localhost:11434"
@@ -34,6 +36,7 @@ class API:
         else:
             return "Error: " + response.text
 
+    # This method sends a message to a certain AI.
     def send_message(self, message, model):
         if model == 1:
             answer = self.process_text_local(message, "phi3.5")

py/venv.sh

@@ -5,3 +5,6 @@ source venv/bin/activate
 pip install transformers
 pip install torch
 pip install flask
+pip install SpeechRecognition
+pip install pyaudio
+pip install pocketsphinx

py/web_flask.py

@@ -1,11 +1,15 @@
 from flask import Flask, request, render_template
 from api import API
+from voice_recognition import Voice
 
 APP = Flask(__name__)
 api = API()
+voice = Voice()
+
 messages = []
 
 
+# The following method shows the user the GUI and does the backend connections to the API.
 @APP.route('/', methods=['GET', 'POST'])
 def index():
     global messages
@@ -16,6 +20,7 @@ def index():
         user_message = request.form['user_message']
 
         if option == "voice":
+            user_message = voice.listen()
             messages.append(f"User: {user_message}")
         elif option == "chat":
             messages.append(f"User: {user_message}")