"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

ReSpeaker Lite Voice Assistant Kit, ESPHome Assistant, 2 Mic Array with XIAO ESP32S3 pre-soldered, Mono Enclosed Speaker from Aliexpress (https://www.aliexpress.com/item/1005007469752193.html?spm=a2g0o.order_list.order_list_main.11.5cf51802fhyBvV)

I don’t see any errors in my dashboard…
Your log shows nothing about attempts to recognise the wake word, or buttons clicks.

Try clicking action button please. Also verify your HA pipeline, and try playing some music on it, or TTS announcement.

Did you flash the ReSpeaker Lite (not the ESP32S3)?

@formatBCE think it would be hard to incorperate temperature sensor, and light lux sensor in your setup, be a nice add-on even closer to the features coming in the futureproofhomes satelile1 - even the presence sensor would be an absolute cherry on top for an all-in-one satellite.

Hmm, I don’t like that idea. Let me explain:

  1. SAT1 has custom PCB, so it has almost everything onboard. With DIY solution like Koala it would be cords mess.
  2. I don’t want to overcomplicate the setup. Adding bunch of YAML for sensors will make debugging much harder.
  3. I have SAT1, and I know how much effort (with custom modules and stuff) they made to get presence sensors working in parallel with voice. And still they fight some occasional glitches. S3 is powerful, but not powerful enough to hold too much.
  4. Temp/humidity sensors inside of enclosure will suffer from non-linear aberration, and it will be pretty hard to predict correction. Also, often the place for voice assistant and place for effective temp/humidity readings are different places. Not to mention that presence sensor should be placed precisely and in persistent position, while Koala can be moved easily, defeating purpose of presence sensor zones etc.
  5. Since S3 boards are so cheap, I prefer having eggs in separate baskets myself. I have separate presence sensors, ambient quality sensors etc on different devices, with tiny footprint or hidden completely. Effective management, simple configuration - and also if one of my devices does, it’s just single sensor instead of important combine.
2 Likes

From the log, it’s 1.1.0, so flashed :slight_smile:

Makes sense, after tonight I should have all 4 devices operating to start on my YT video for comparison on all options out there minus the sat1 which is a waiting list at this point. Wyoming-Satellite with Pi, the voice pe, the ReSpeaker Lite with S3 soldered and the ReSpeaker Lite with S3, with seperate volume. So many options I like the Koala-Satellite option best so far, just wish I found it first I’ve spent weeks on testing all these different options and waiting for parts.

1 Like

Yeah, I feel like it deserves more visibility. I keep finding same questions about voice assistants everywhere…

1 Like

So true and there are so many variations. Can you clarify for me who is behind most of the Voice Assistant work for wyoming-satellite, PE, SAT1, Koala, etc. there are so many branches. And I want to be giving credit where credit is due. :slight_smile:

Voice PE is created by NabuCasa developers. It’s the reference for every other project on ESP chips now.
SAT1 is project of FutureProofHomes. They took PE firmware, and extended it (and hardware too!) to the maximum. Brad is really cool guy, and i’m there in initiative group too.
Respeaker (and Koala in particular) is something i started myself, basing on PE firmware, and putting all the pieces together (like resampler for microphone, created by community member, 48-kHz firmware for Respeaker itself, and custom component for Respeaker ESPHome integration, that i made (mostly by copying and adapting the code from VPE project too).

Wyoming is standing a bit apart here - it’s basically just protocol. Wyoming-satellite on Raspberry with 2-mic hat is great foundation, but it’s too complicated for me to contribute, and too restricted in features to use it without modifications. At least for me, Koala works much, much better regarding stability, feature set and ease of modifications.

1 Like

Awesome thanks! Yeah I started with Wyoming and wow what a chore to get going and working. Did you design the case for Koala yourself too?

Yes I did. Many iterations, and mediocre final design - but it fits my interior. :slight_smile:

Yeah, I started with Wyoming too. It tortured me for several months, until I decided to put it in the drawer (even with FutureProofHomes enhancements it still sucked)…

My Wyoming works fine, audio is a little crackly but I’m sure that is one of my settings. I have it side by side the other ones and it works just the same other than the cool updates and extra controls you can do right from HA.

Well done, not sure if it’s my printer but all the pieces is well over 30 hours of printing and I need to build 8 of these satellites, might design a smaller case :slight_smile:

Oh!
I have Bambulab A1, it’s little over 8 hours for everything.

1 Like

FYI, of course it will a community effort in the long turn but up until now it has been Nabu Casa employeed developers (paid by Home Assistent Cloud subscribers) that has been doing most of the underlying groundwork on behalf of the Open Home Foundation so they are they driving force that is makibg all this happen at this pace (so first read Thinking Bigger: State of the Open Home 2024). And for more of the story do recommend read all their original year of the voice blogs and all the following voice chapter # series of blog posts, etc. which lead up to the the era of open voice assistant. Much but not all of the backstory is summerized or at least linked in these two blog posts, see:

And

Voice control for Home Assistant goes back further than most people assume, with some of the groundwork we use today being added as far back as 2017. The major turning point came when we refocused our efforts and declared 2023 the Year of the Voice. This was an effort to focus development and find areas where our community could make the most impact. During the Year of the Voice Assist was added to voice, intents were improved, languages added, wake words were created, and we established great local and cloud options for running voice. Shortly after Year of the Voice many more features were added, including integrated AI, timers, and even better wake words. Year of the Voice got the ball rolling, and Voice Preview Edition will continue its momentum.

2 Likes

Anyone have an image of this? Where is the In-Out?

  • Bridge IN-OUT and RGB with solder on ESP board. (Hint: IN-OUT enables 5V pin, and RGB pad makes onboard LED working. You may skip latter, if you don’t plan to use LED on th eboard, but don’t skip IN-OUT pad, unless you’re sure that 5V pin on board provides current.)

Take a picture of your ESP32-S3. Probably you don’t have that…?
P.S. you may don’t have that. Sorry it’s unclear from README…
It all is basically just to make sure that 5V pin and LED are alive. On some boards it’s optional.

I measured pin it has 5V but RGB LED on ReSpeaker Lite is not working.

I guess there’s confusion with LEDs. :slight_smile:
You’re good with 5V. That LED I mention is ESP LED, not the Respeaker LED.

If you solder Respeaker correctly to ESP, LED on it will be working.