robmsmt

sounds strange, I would have expected a Java Android app with jni calls to c/c++ through Android ndk

Photopea is excellent. thank you sincerely.

Regards oftentimes forgotten Linux user

Thanks, this was my thinking. I took a look at microG and couldn't see any reference to it on the status page: https://github.com/microg/GmsCore/wiki/Implementation-Status

The options to make a true opensource assistant that would work in the same capacity as Alexa/Siri/Google (invoked via wakeword) is therefore quite limited. An alternative is to have it listening constantly, maybe as a background service, but this would use up too much battery I suspect.

>To choose Dicio as your default digital assistant on Android (replacing Google Assistant), go to Settings > Apps & notifications > Default apps > Digital assistant app > Default digital assistant app. On Android 10+, this lets you activate Dicio by swiping from the bottom left/right corner toward the center of the screen.

This is great, one of the barriers to entry before was that you needed to manually open the app. Although I don't see that this resolves the issue of the wakeword/hotword triggered with "Hey Google" ?

I am sorry to hear this. I think there are many people in a similar boat to you and there are quite a few people working on command & dictation computing. Although my tool _may_ help you find out which speech systems work well for your voice/accent/mic/vocab it might also be worth trying another one of the specialist libraries specifically for dictation and controlling computers.

I've not heard of Almond, but I have seen the following projects which might be helpful:

- Dragonfly: https://github.com/dictation-toolbox/dragonfly

- Demo: https://www.youtube.com/watch?v=Qk1mGbIJx3s / Software: https://github.com/daanzu/kaldi-active-grammar

Far field audio is usually harder for any speech system to get correct, so having a good quality mic and using it nearby will _usually_ help with the transcription quality. As a long time Linux user, I would love to see it get some more powerful voice tools - really hope that this opens up over the next few years. Feel free to drop me an email (on my profile) happy to help with setup on any of the above.

I created a toolkit to evaluate many different speech recognition engines.

https://github.com/robmsmt/SpeechLoop

Comparing speech systems can take a long time esp for a dev who doesn't have the background in audio/ml. How do you know which one will work best? Will new shiny transformer model perform well enough? Most end up using one of the big tech companies existing API to throw their data at. Whilst this is convenient, I think that it's a travesty that opensource speech systems have not are not as easy to use. I was hoping to change that to make it easy to evaluate and compare them!

i did a quick comparison with regards to word error rate... https://robmsmt.github.io/2021/09/04/benchmarking-asr-first-...

For my particular dataset it worked quite well. It was the best offline ASR that I tried.

your company sounds like it would be very useful to many startups. The tooling stack we went for is:

Jira+Confluence, circle CI as the CI/CD service Artifactory for binary hosting and GitHub+gitlfs

circle had the lowest friction compared to Jenkins and GitHub actions wasn't out at the time!

Thanks it looks neat. I'll check out your scripts - especially interested in the power draw part for myself. I had setup the 80% cycling before but that was on my n-1 installation and haven't had chance to put that back on.

I have a friend who I convinced to buy a Thinkpad and stick ubuntu 20.04 on it. Unfortunately he got a more expensive X1 one with an Nvidia Graphics card, it strangely overheats when he needs to charge it and plugs it into the AC. I suspect the power settings are all over the place. I've not had chance to get a log from him yet. How did you "power down" your nvidia card?

if you could point to how you configured your power management that would be great

a way that I've found to do this securely is to have a small constantly running cloud server $5 pm with OpenVPN on it. Then have your server connect and join the VPN on both laptop and server. If it's configured properly you'll be able to get them to communicate

I am working on something to compare at least 10 different ASRs both open source and production ones.

I think we as a community need to come up with officially recognized names for such activities to lessen the cognitive burden on installing such software...

You can also self host with GOGS / Gitea and no limit but getting my company to move from Github will be a large undertaking. It's not worth it for GitLFS on it's own.

Pushing Github past the 100mb limit has to be the most requested feature. Ridiculous that we have to use the fudge that is GitLFS.

It just adds complication for a limit that shouldn't be there anyway.

it reminds me when I stumbled upon the fast_api docs recently. I have to admit I found it more confusing. https://fastapi.tiangolo.com/async/#asynchronous-code

Glad this tool exists I have just moved to it. The main thing missing in my opinion to record a video as a gif.

I also agree. Sometimes I speak to work colleagues who are on BT headsets and it must add what feels like 100ms to the call. I can sometimes hear the end of my sentence as I speak. Infuriating.

You sound like me with every product i try to build.

For those of us doing any ML that requires CUDA/CuDNN. It feels that AMD (specifically for GPU) still isn't a viable option at this moment in time. I sincerely hope that this changes.

I meant to say that these comments from Bob Martin remind me of testing and proper architecture for the code

Isn't it against the ToS to have multiple accounts? They might ignore these accounts for now, but in the future?

We mostly use gitflow. Commits do not just "sneak in". Master is locked and needs a PR/CR to merge. The process usually takes care of the majority of the issues and usually works okay for us.

3rded... wait maybe there's an idea here.

McPaywall

A better idea would be for hacktoberfest to send everyone who has made a 1-line change to a README.md as their PRs to be given a dunces hat instead.

I don't understand why a person or persons create new repos and merge PRs into each others hello world empty repos. This would be be the optimal solution no?

Next week... Chrome TV+

First thing I noticed when I moved to the US. It was quite the shock.