Exploring Microphone Arrays In Robotics
I've been exploring the use of microphone arrays in robotics, and this video provides some insightful information on the topic. Like many projects I've undertaken, I started this with great excitement. However, as I've delved deeper, I've found the challenges to be more complex than I initially anticipated.
Despite the hurdles, the enhancements in the system's robustness and responsiveness have been significant, making the effort worthwhile. My main objective is to achieve effective audio head tracking or, at the very least, accurate noise localization.
I plan to continue sharing my progress and insights along the way, and I hope to eventually share a success story. I'd also appreciate hearing about others' experiences with microphone arrays in robotics. If you've worked on something similar, please share your journey!
So, the mic array is now installed and working quite satisfactorily. Eye and head movements can be mixed quite elegantly using a mixer (ARC skill Relative Servos). This makes the movement towards the audio source look very realistic. However, the SBC's fan noise in the upper body still needs better filtering. There's room for improvement there. But for now, it's Christmas.
We'll continue next year. I wish the community a Merry Christmas and a Happy New Year.
Robo-Chess
-
That's such a nice calming video - thanks, i needed that today! the microphone array looks cool and i'm glad it works. i don't have any experience with microphone arrays but oneday hope to try them. i really like the displays on the robot, looks very modern.
have you ever tried playing with the azure text to speech robot skill? you don't need to change any script commands if you add it. it'll automatically take over any Say() commands if this is checked. i find there's a few voices that have a decent robot-ish style to them.
thanks again for sharing that video! happy holiday to you and your robot family as well!
Thank you very much.
Yes, I'm familiar with the Azure text-to-speech robot skill, and it's the best TTS option for me.
However, in the video, the robot is speaking via an external TTS app from ARC. The reason is that GPT is telling the story and simultaneously controlling the robot's movements. The separation of text and control commands is handled by a Python program. The movement commands are routed directly to ARC via a virtual COM port, triggering actions there. That's why the robot is speaking through Windows. It doesn't sound nearly as good as when using Azure. Unfortunately, all attempts to have ARC execute the text and movement on the fly have failed due to timing issues.