/interactive-azure-openai-speech-chat

Primary LanguagePython

interactive-azure-openai-speech-chat

This is a simple example that uses Azure Speech and Azure OpenAI Service to implement a chat-like experience using voice.

Copy .env.example to .env, and adapt the values:

OPENAI_BASE_URL=https://xxxxxxx.openai.azure.com/ # point to your Azure OpenAI Service URL
OPENAI_API_KEY=xxxxx # fill in your Azure OpenAI Service API key
OPENAI_DEPLOYMENT_NAME=text-davinci-003 # fill in your text-davinci-003 deployment name
SPEECH_API_KEY=xxxx # fill in your Speech API key
SPEECH_REGION=westeurope # adapt to your Speech API Azure region

# optional parameters
SPEECH_VOICE_NAME=de-DE-AmalaNeural # optional, defaults to de-DE-AmalaNeural
LOCALE=de-DE # optional, defaults to de-DE
INCLUDE_HISTORY=True # optional, defaults to True, which includes history in prompts for follow up questions

Then install the dependencies and run the script:

pip install -r requirements.txt
python sample.py

UI support

If you want a simple web interface to visualize the chat conversation, run the included streamlit UI:

streamlit run sample.py

Please not that the STT/TTS does not run client-side, so this application can only run on localhost.