"ReSpeaker Lite" - new Seeed Studio Voice Assistant Development Kit hardware combine ESP32 with XMOS XU316 DSP chip for advanced audio processing as a ESPHome-based Home Assistant Assist Satellite voice devkit

I would love to give this new enclosure a try! Could you publish this as a work in progress?

1 Like

Is this 3d printed? Any stl files available publicly?

It is 3d-printed.

I didn’t yet publish STL/STEP, since i’m changing it a lot. Also it presumes that you’re using Devkit-C board, not XIAO, and also additional LED strip and rotary encoder. So i need to make instructions for soldering and assembly.

Hi there,
just a short question before ordering one of these; I played around with ESP S3 and INMP441 before, worked well when directly facing the device. But even turning away from it while being real close resulted in really poor audio and no recognition. What would you estimate the range that the Respeaker is good at understanding stuff? Just to get an idea.
THX in advance
Unlight

It works at least up to 20 feet 75% of the time. Probably 100% 12 to 15ft (3 meters) away. That is if no TV is on, it can pick that up and start listening once you have triggered it with the wake work. It’s good with background noises like an air conditioner or music. That’s the only minor issue I have and I just wrote an automation to lower my soundbar really low then raise the volume back up once it’s done as a workaround until the get that figured out, which I am sure they will eventually but that one sounds extra fun.

Newest firmware is also 48000Khs, before it was 16000Khz so sounds better also now. At least for voice.I can’t really tell if it helps on the microphone side or not but one would think it would be clearer to convert from SST to process.

1 Like

Ok, thanks. I already had an automation for the exact same scenario, lowering the volume if music is playing. I think I’ll give it a try.

I’m interested in buying a respeaker lite to evaluate echo cancellation and VAD - I’m not particularly interested in activation via wake word at the moment, just the ability to deliver clean audio files to my application even while playing speech via a speaker.

Is the need to hear a wake word hard-wired, or can it just listen and record continuously?

(I presume you meant STT not SST?)

You mean with ESPHome?

Respeaker can work as USB microphone with PC/Mac, using USB firmware.

If you want to use it with ESPHome, then i guess you can just build UDP player with it and stream microphones wherever you want to.

1 Like

At this point, unless you want to use the USB port for something like the “assist microphone” add on then I would wait until Dec 19th after HA’s Livestream on their hardware which will have an XMOS chip also. I believe it’s the same model XMOS chip but someone correct me if I’m mistaken. I don’t think it has a USB firmware like the respeaker lite does but I would wait for the live stream for confirmation on that. If you’re interviewing I would purchase as fast as possible after it goes on sale. While Nabu Casa/HA has said they have learned a lot from previous hardware sells, demand could be high. I pre-ordered the Skyconnect when it was announced and it was a nightmare of delays but that was also their first hardware product sold.

While I’m surprised and grateful Seeed studio actually had firmware updates and support for the respeaker lite, I imagine the HA team has more time and resources to work on making their VA better.

Honestly, the biggest announcement of the year has yet to come though… :shushing_face: I’m pretty sure it is voice hardware related. :wink: So make sure you aren’t missing the live stream on 19 December! :tv: You will not be disappointed! :gift:

Case is, that I use Nabu firmware for Respeaker. For sure, XMOS DFU software can be better - but ESPHome wise it’s identical. And yes, you’re right - it’s same chip XU316.
But yeah, for end users PE will be to-go device.

1 Like

Also don’t forget about SAT1 from FutureProofHomes. It should be awesome. :wink:

2 Likes