Hey everyone! I’m trying to set up a cool project but I’m stuck. I want to ask my Google Home to talk to ChatGPT through n8n. The idea is to say “Hey Google, ask GPT” and then get a spoken answer from ChatGPT.
I’ve got n8n running with a workflow, an OpenAI key, and a Google cloud account. I’ve done some setup in the Actions console and got OAuth2 stuff going in n8n. But when I try to put the OAuth2 ID and secret into Google, it asks for a scope and I’m lost.
My questions are:
- What scope should I use?
- How do I make n8n work with Google Home voice queries?
Also, is this even doable now that Dialogflow is gone? I can’t find any up-to-date guides. If anyone has done this or knows how, I’d really appreciate some help!
Thanks!
I’ve actually attempted a similar setup, and it’s quite challenging without Dialogflow. The scope you’re looking for is likely ‘https://www.googleapis.com/auth/assistant-sdk-prototype’, but that’s just part of the puzzle. The real hurdle is interfacing with Google Home’s voice capabilities directly through n8n. From my experience, you might need to consider using a middleware solution. Something like IFTTT or Home Assistant could potentially bridge the gap between Google Home and n8n. Alternatively, you could explore using Google’s Actions SDK, but it’s more complex and requires coding skills. It might be worth investigating if you’re up for a deeper technical challenge. Remember, Google’s policies on third-party voice interactions are quite strict, so ensure you’re complying with their terms of service throughout your project.
I’ve tackled a similar project, and it’s definitely a complex undertaking. One approach that worked for me was using Node-RED as an intermediary between Google Home and n8n. It has nodes specifically for Google Assistant integration, which can handle the voice input. From there, you can set up an HTTP request to your n8n workflow.
For the OAuth scope, try ‘https://www.googleapis.com/auth/assistant-sdk-prototype’ as mentioned earlier. However, be prepared for some trial and error. The tricky part is getting Google Home to recognize and route your custom commands correctly.
A word of caution: this setup can be quite unstable and prone to breaking with updates. I found myself tweaking and fixing things frequently. If you’re open to alternatives, you might want to look into using a Raspberry Pi with a microphone as a DIY smart speaker. It’s more work upfront, but gives you much more control and flexibility in the long run.