nardial 0.1.3

Nardial Dialog System

NarDialPy

Python implementation of the Narrative Dialog Management System.

---

Demo

Here is a demo showcasing an agent-driven conversation utilizing Google Dialogflow, Google TTS, and OpenAI's GPT4

First, you need to set-up Google Cloud Console with dialogflow and Google TTS:

1. Dialogflow: https://socialrobotics.atlassian.net/wiki/spaces/CBSR/pages/2205155343/Getting+a+google+dialogflow+key
2. TTS: https://console.cloud.google.com/apis/api/texttospeech.googleapis.com/ Note: you need to set-up a paid account with a credit card. You get $300,- free tokens, which is more then enough for testing this agent. So in practice it will not cost anything.
3. Create a keyfile as instructed in (1) and save it conf/dialogflow/google_keyfile.json (Never share the keyfile online!)

Secondly you need to configure your dialogflow agent. 4. In your empty dialogflow agent do the following things:

• Remove all default intents
• Go to settings -> import and export -> and import the resources/droomrobot_dialogflow_agent.zip into your dialogflow agent. That gives all the necessary intents and entities that are part of this example (and many more)

Thirdly, you need an openAI key: 5. Generate your personal openai api key here: https://platform.openai.com/api-keys 6. Either add your openai key to your systems variables or create a .openai_env file in the conf/openai folder and add your key there like this: OPENAI_API_KEY="your key"

Forth, the redis server, Dialogflow, Google TTS and OpenAI gpt service need to be running:

7. Run:

pip install --upgrade social-interaction-cloud[dialogflow,google-tts,openai-gpt]

8. Run:

conf/redis/redis-server.exe conf/redis/redis.conf

9. Run in new terminal:

run-dialogflow 

10. Run in new terminal:

run-google-tts

11. Run in new terminal:

run-gpt

12. Connect a device e.g. desktop, nao, pepper, alphamini
13. Run this script in a new terminal.

---