voices added

app.py

@@ -3,19 +3,48 @@ import openai
 import pandas as pd
 import numpy as np
 import gradio as gr
-import openai, subprocess, os
+import openai, os, re
 import requests, base64
 from io import BytesIO
 from huggingface_hub import Repository
 from utilities import api_keys, prompts
 from create_embedding import QueryEmbeddings
 
+import pkg_resources
+pyttsx3_version = pkg_resources.get_distribution("pyttsx3").version
+print(f"pyttsx3version: {pyttsx3_version}")
+
 openai.api_key = api_keys.APIKeys().get_key('OPENAI_API_KEY')
+eleven_api_key = api_keys.APIKeys().get_key('ELEVEN_LABS_API_KEY')
+voice_id1 = api_keys.APIKeys().get_key('VOICE_ID1')
+voice_id2 = api_keys.APIKeys().get_key('VOICE_ID2')
+voice_id3 = api_keys.APIKeys().get_key('VOICE_ID3')
+voice_id4 = api_keys.APIKeys().get_key('VOICE_ID4')
+voice_id5 = api_keys.APIKeys().get_key('VOICE_ID5')
+voice_id6 = api_keys.APIKeys().get_key('VOICE_ID6')
+voice_id7 = api_keys.APIKeys().get_key('VOICE_ID7')
 
+voice_id = voice_id1
 messages = [ {"role": "system", "content": "**hello**." }]
 
-def transcribe(audio):
-    global messages
+def Transcribe(audio, voice):
+    global messages, voice_id
+    print ("voice: " + voice)
+    if voice=="Voice 1":
+        voice_id = voice_id1
+    elif voice=="Voice 2":
+        voice_id=voice_id2
+    elif voice=="Voice 3":
+        voice_id=voice_id3
+    elif voice=="Voice 4":
+        voice_id=voice_id4
+    elif voice=="Voice 5":
+        voice_id=voice_id5
+    elif voice=="Voice 6":
+        voice_id=voice_id6
+    else:
+        voice_id=voice_id7
+        
     audio_filename_with_extension = audio + '.wav'
     os.rename(audio, audio_filename_with_extension)
     
@@ -30,18 +59,41 @@ def transcribe(audio):
 
     response = openai.ChatCompletion.create(model="gpt-3.5-turbo", messages=messages)
     system_message = response["choices"][0]["message"]
-    messages.append(system_message)
-
-    subprocess.call(["say", system_message['content']])
+   # messages.append(system_message)
+    
+     # text to speech request with eleven labs
+    url = f"https://api.elevenlabs.io/v1/text-to-speech/{voice_id}/stream"
+    data = {
+        "text": system_message["content"].replace('"', '').replace('401k', '4 oh 1 k'),
+        "voice_settings": {
+            "stability": 0.95,
+            "similarity_boost": 0.93
+        }
+    }    
+    r = requests.post(url, headers={'xi-api-key':eleven_api_key}, json=data)
+    r_audio =r.content
+    audio_io = BytesIO(r_audio)
+    audio_io.seek(0)
+    audio_base64 = base64.b64encode(audio_io.read()).decode("utf-8")
+    audio_html = f'<audio src="data:audio/aac;base64,{audio_base64}" controls autoplay playsinline></audio>'
+    return system_message['content'], audio_html
 
-    '''chat_transcript = ""
-    for message in messages:
-        if message['role'] != 'system':
-            chat_transcript += message['role'] + ": " + message['content'] + "\n\n"
-    print(chat_transcript)'''
+#ui = gr.Interface(title="USAA Benefit Advisor", fn=transcribe, inputs=gr.Audio(label="Question", source="microphone", type="filepath"), outputs=[gr.Textbox(label='Response'),gr.HTML()]).launch()
+#ui.launch()
 
-    # return chat_transcript
-    return system_message['content']
+'''comp = gr.Interface(
+    Transcribe,
+    title="USAA Benefit Advisor", 
+    inputs=[gr.Audio(label="Question", source="microphone", type="filepath")], 
+    outputs=[gr.Textbox(label='Response'),gr.HTML()])
+'''
+comp = gr.Interface(
+    Transcribe,
+    title="USAA Benefit Advisor", 
+    inputs=[gr.Audio(label="Question", source="microphone", type="filepath"),
+            gr.Radio(["Voice 1", "Voice 2", "Voice 3","Voice 4", "Voice 5", "Voice 6", "Voice 7"], value="Voice 1", label="Change voice", info="Which voice would you like to use?")], 
+    outputs=[gr.Textbox(label='Response'),
+             gr.HTML()]) 
 
-ui = gr.Interface(title="USAA Benefit Advisor", fn=transcribe, inputs=gr.Audio(label="Question", source="microphone", type="filepath"), outputs=gr.Textbox(label='Response')).launch()
-ui.launch()
+#demo = gr.Parallel(comp, ui1, title="USAA Benefit Advisor")
+comp.launch()

utilities/prompts.py

@@ -1 +1 @@
-ADVISOR_CUSTOM_PROMPT= "You are an HR benefits advisor for USAA. Don't say your an AI language model. Provide answers with kindness and respect for military personnel and families. Be professional and warm." 
+ADVISOR_CUSTOM_PROMPT= "You are an HR benefits advisor for USAA. Don't say your an AI language model. Provide answers with kindness and respect for military personnel and families. Be professional and warm, keep answers short and sweet."