Arcx 2026.09.30.03

startup log.txt I asked chatgpt to look at the log and tell me if there were any audio issues, the reply: Yes - there are audio-related PortAudio errors during ARCx startup, but they do not appear to be fatal audio failures. ARCx tests the available Windows audio interfaces and rejects several because they cannot handle the exact format ARCx requests. For example:

• Realtek headphones/speakers using WASAPI reject mono 22,050 Hz with PortAudio -9997: Invalid sample rate.
• Logitech BRIO and Realtek microphones using WASAPI reject mono 16,000 Hz with the same -9997.
• The reSpeaker Flex XVF3800 WASAPI microphone rejects the requested configuration with PortAudio -9998: Invalid number of channels. Pasted text

However, immediately afterward ARCx reports:

**33 audio devices found, of which 10 can play and 10 can record in the format ARCx uses.**

Then it successfully opens:

* **Playback:** `Headphones (Realtek(R) Audio) - MME`, device 5
* **Microphone:** `Echo Cancelling Speakerphone (reSpeaker Flex XVF3800...) - MME`, device 1

And finally:

**Audio stream started: separate streams for recording and playback**.
``` Pasted text

So I would classify this as **ARCx probing incompatible versions of your audio devices, then successfully falling back to compatible MME devices**.
Most importantly, I don't see an actual audio startup failure after that. **Your reSpeaker microphone did initialize successfully**, just through **MME rather than WASAPI**.

I then went into Options and Preferences in ARCx. the audio devices were set to system default. I went in and changed to the below:

and the good news, Sound is working without error. Hopefully this version has fixed the issue. But wow 33 audio devices, not sure where they are came from.

Thanks. will let you know if i run into errors.

If I could, I would like to put the Brushless Motor Panel and the ARCx skill request list. Thank you.

Haha, oh boy, ChatGPT response but it isn't wrong. The startup log message from the last ARCx is indeed filtering out audio devices that don't match the desired sample rate. We had lengthy conversations and analysis of several years of data to determine how to optimize ARCx moving forward. One thing we determined is that the robot's computer does a lot of heavy lifting. The idea of ARCx is to not only offload the GUI to a remote computer's web browser, but also limit unnecessary processing and storage. One of those is audio - there's no need to be playing back 44khz or 128khz audio, haha - that's ridiculous. The same applies to audio recording, which is only used for speech recognition. There's no need to exceed the audio recording sample rate when speech recognition uses such a low rate. So the point is, you're seeing the log file of ARCx, which offers a ton of transparency to understand what's going on. And using ChatGPT might be a good source to help add clarity to the log noise. Athena can, too, if needed.

We will add the brushless motor panel to ARCx for you - stay tuned.

We are prioritizing robot skill development with feedback from our testing group and the few of you in the community who have mentioned it. Mostly, just you So thank you, and we'll keep pushing them out based on that feedback.

We have a bunch of new robot skills coming out either tonight or tomorrow.

Thank you for that. I am running ARCx exclusively and it seems faster and more steady for my use. I have around 60 individual scripts, most of them are small, like reading if an EZB is connected or not. But a few are a little more involved, including a scheduler that runs all the time and polls the little scripts, getting data, speaking a status or two, etc. With ARC I was getting some out of memory errors and a few other errors, but not now. Also, I have many MP3 files and I mean many. Also, now that I'm used to it, I like the script interface.

also a cosmetic fix request. when project description does fit in box it just overruns the box.

Can you tell us what that screenshot is from? For example, is it a robot skill page, project page, etc..

The home page.