If you are blessed with AMD card and wanna play with machine learning tools well... then you are screwed because #ROCm is a pure crap at leas on Linux. AMD really rocks with everything else on Linux desktop but not with ml.
I was looking for #flux toolset that could run with llama.cpp and I found one: https://github.com/leejet/stable-diffusion.cpp
It can run #flux and #stablediffusion with #Vulkan as a backend. Really impressive.
I'm fascinated by how often AI autocomplete is both right and wrong.
It will often autocomplete the thing I intended, and it makes me say "oh good, I already implemented that function"
But then the IDE tells me there's no function by that name, and I realize the AI completed what I *wanted* not what I *had*.
I'm used to expecting the IDE to only autocomplete what actually exists.
I just watched and really enjoyed #38c3 talk "io_uring, eBPF, XDP and AF_XDP".
https://media.ccc.de/v/38c3-iouring-ebpf-xdp-and-afxdp
The presenter showed some techniques to 10x speed up network processing in Linux.
Very informative :)
#SpeechNote 4.7.0 is now available on #Flathub!
Release highlights:
- Vulkan GPU acceleration (much faster STT with Intel, AMD or NVIDIA graphics cards)
- Rules for text transformations after STT or before TTS
- Echo mode (after STT text will be immediately read out)
- Many new languages supported in Translator
Video presenting all the changes: https://youtu.be/cEht4Fts6Bo
Full changelog: https://github.com/mkiol/dsnote/releases/tag/v4.7.0
Remember: IDN is crazy. And just not just a little.
I've just released version 0.3 of Pied (https://pied.mikeasoft.com), my tool for making advanced text-to-speech voices work easily on Linux.
This release adds support for models with multiple sub-voices, providing access to hundreds of new voices. It also adds different qualities of model, fixes previews whilst downloading and fixes audio output detection when using sox.
#TextToSpeech #TTS #ScreenReader #Linux #Flutter
Today we have a really fun one, the developer of ZLUDA, the project bringing CUDA support to non-Nvidia GPUS, you know the one that AMD paid for and took down. #Nvidia #CUDA #AI
Video: https://www.youtube.com/watch?v=ze25Sie2gVQ
Audio: https://open.spotify.com/episode/764QsXmqR6g2ITxaY1TEHn
@revanmj ooo, to nawet działa lepiej niż się spodziewałem! To taka twoja sekretarka pytająca "a w jakiej sprawie?"
@revanmj "zsyłam na bota" - Dobrze mieć bota po swojej stronie :) Jedne z nielicznych zastosowań IA które akceptuję...
💾 Software-Tipp: Speech Note
Unfassbar gute Diktier-, Vorlese- und Übersetzungs-Software für Linux und Sailfish OS. Im Hintergrund läuft das, was die KI-Entwicklung heute kann - aber lokal, ohne Cloud, Lizenzen, Datenabfluss oder dergleichen. Man kann zwischen leistungsstarken, ressourcenhungrigen und kleineren Modellen wählen. Installation für Technik-Dummies wie mich, als Flatpak. Entdeckt in der aktuellen c't von @heiseonline.
1000 thanks, @mkiol. 🙏
@Lambo @heiseonline You're welcome. I'm glad it's useful :)
@darkhog this is definitely possible. I'm way too lazy to create a new app, but perhaps I could add something to Speech Note to support this use case. I will think about it :)
@darkhog do you have in mind a way to control the computer with your voice? For example, you say “open notepad” or “shutdown the system”?
[$] Transcribing audio with AI using Speech Note
One of the joys of writing about technology is the opportunity to cover interesting talks on open‑source and free‑software topics. One of the pains is creating transcri [...]
🇪🇺⚖️ Today is a good day for European citizens, tax justice, and fair competition.
In its final judgment, @Curia has confirmed our 2016 decision that Ireland granted illegal aid to Apple, requiring the release of up to 13 billion euros in unpaid taxes.
The court has also upheld our decision against Google for abusing its dominant position by favouring its own comparison shopping service and the resulting €2.4 billion fine.
ℹ️ https://europa.eu/!Dy3ndR
#EU
South Pacific Shadowset
Image Credit & Copyright: Jin Wang
In my honest opinion it is actually more like this:
If you experienced a crash when starting #SpeechNote 4.6.0, I just released an update with a fix.
New version 4.6.1 is now available on #Flathub!
Update also includes:
- new Translator models (Danish, Croatian, Slovenian, Indonesian, Romanian)
- fixes for handling special keyboard keys as keyboard shortcuts (XFree86)
- updated Swedish translation