randkyp

Neat! While the physicality of having the CD spin while running inference is undeniably cool, I wonder if you could run larger models at higher speeds through the PS2 HDD accessory/Memory Card Micro SD adapter/the PS2's USB port.

I doubt the VUs can help with inference given their small scratchpad sizes and instruction set though, haha.

It's a trademark thing; there are some prerequisites and manual approval involved before you get to officially name your distro "[Distro Name] Linux".

https://www.linuxfoundation.org/legal/the-linux-mark

Heh, funny you mention that, considering Be's pivot to BeIA. Some Be engineers also worked on the (unreleased) Palm OS Cobalt, and eventually, Android. (And then Fuschia, but I don't think that OS will ever hit smartphones.)

Haiku does support WLAN adapters (even USB ones), through the support isn't as extensive as Linux or the BSDs. You might want to use the current nightly builds instead of the latest beta version, though, which was released in December 2022.

https://hardware.besly.de/index.php?hardware=Wireless_Networ... https://discuss.haiku-os.org/t/support-for-realtek-usb-wifi-...

Does this do RAG over the character's chat history too? That's something SillyTavern can also do with extensions, but I figured since your project already uses Llamaindex, this feature can be something that's already baked in from the get-go.

I played with it some more and I have to agree. For actual voice _cloning_, XTTS2 sounds much, much closer to the original speaker. But the resulting output is also much more unpredictable and sometimes downright glitchy compared to OpenVoice. XTTS2 also tries to "act out" the implied emotion/tone/pitch/cadence in the input text, for better or worse.

But my use case is just to have a nice-sounding local TTS engine, and current text-to-phoneme conversion quirks aside, OpenVoice seems promising. It's fast, too.

This is HN, so I'm surprised that no one in the comments section has run this locally. :)

Following the instructions in their repo (and moving the checkpoints/ and resources/ folder into the "nested" openvoice subfolder), I managed to get the Gradio demo running. Simple enough.

It appears to be quicker than XTTS2 on my machine (RTX 3090), and utilizes approximately 1.5GB of VRAM. The Gradio demo is limited to 200 characters, perhaps for resource usage concerns, but it seems to run at around 8x realtime (8 seconds of speech for about 1 second of processing time.)

EDIT: patched the Gradio demo for longer text; it's way faster than that. One minute of speech only took ~4 seconds to render. Default voice sample, reading this very comment: https://voca.ro/18JIHDs4vI1v I had to write out acronyms -- XTTS2 to "ex tee tee ess two", for example.

The voice clarity is better than XTTS2, too, but the speech can sound a bit stilted and, well, robotic/TTS-esque compared to it. The cloning consistency is definitely a step above XTTS2 in my experience -- XTTS2 would sometimes have random pitch shifts or plosives/babble in the middle of speech.

I found this recent thread interesting, specifically about really considering whether you're going to read the data you just wrote in the near future or not (in which case, use direct IO) and a set of (abandoned?) patches for write-behind caching for sequential writes in Linux (https://lore.kernel.org/lkml/156896493723.4334.1334048120714...).

https://lore.kernel.org/linux-mm/45odvhgymm7fxsgwpewoiiggaok...

The unlock tool would only work if it successfully authenticates with Xiaomi's server with matching Mi Cloud ID as the one previously registered to the device. So I very much doubt that it is stolen.

The image transcription in the alt-text is a nice touch.

"A search is either incremental or excremental."

- Jef Raskin

Glad to see that noise reduction is on the roadmap! Does Filmulator support embedded lens profiles? I enjoyed using Filmulator btw, its default output are just lovely.

Dark Reader is open source, under the MIT License: https://github.com/darkreader/darkreader/blob/master/LICENSE

If the equalization is done near the end of the audio output chain (ex: music player, defaults at 100% volume -> user sets master audio output volume to 50% -> EQ is applied) then a pre-amp is not strictly necessary if the EQ can take advantage of the extra headroom in the output channel. Otherwise, an adjustable pre-amp / gain setting would be nice, preferable with the option for it to automatically set itself to reduce by the maximum gain the EQ is applying to avoid clipping. But yeah, it should be optional; if the user wants stuff to clip, let it clip.

Any way to set pre-amp to avoid clipping or is there built-in headroom / automatic gain adjust so that that is taken care of by default?

Switching to another virtual desktop (Win+Tab) can help if the graphics driver is still working normally. If the driver crashed, you have to wait for TDR to kick in and restart the driver (usually ~5s). That is, if it's able to recover.

I've had better luck killing such "unkillable" processes with taskkill.exe /f /im program.exe instead.

The G203 has onboard memory (which also includes RGB light state / startup effect and DPI steps), so once you have your settings right you can just store the profile in the mouse and uninstall the software.

The latest version of the Logitech software (G HUB) doesn't make this feature obvious though, you have to set a local profile then click the gear icon > my gear > click the mouse image > switch on on-board memory mode > click Slot 1 (Default) > click replace with [the profile you just customized, usually Desktop (default.)

The old Logitech Gaming Software was much simpler in this regard.

You can middle-click on the taskbar icon to do that -- and it works for other apps, too.

As weird as it sounds, it might be related to your system region settings, specifically the decimal point sign and the thousands separator sign. I've been only able to open CSVs by manually importing them with Excel's 'import data from text file' function.

For me, it's both. Nesting is helpful as a visual history indicator and a quick way to prune unused tabs that were opened closely in time. Great when you're doing a quick research. The vertical display helps locating tabs by its title and in general, narrower text on a page is easier to read. (Most screens nowadays are widescreen. Some blogs even overcompensate for this by having huge margins and even huge ads on the sidebar, but that's another story.)

I thought their search was powered by Yahoo's BOSS service combined with their own crawler.

I tried pinging google.com from cmd and it worked, both on Firefox and Chrome. Didn't try browsing the web though. Maybe the VPN (https://news.ycombinator.com/item?id=15278430) is overloaded or blocks that somehow?

There's a preliminary support for formulas with only 5 functions or so, so it's not a fully functional spreadsheet yet.

Yes. I just run f3probe for a quick capacity & speed test and then run a single-pass, random data badblocks test just to be sure. If there are any 'bad' sectors, I just pass the badblocks output to mkfs. Be careful with your choice of card reader/adapter though--some starts to get too hot and end up breaking the card under heavy testing!

This feature comes by default with PulseAudio, maybe Ubuntu doesn't expose it well enough in their audio settings. I think Gnome Settings has it, KDE definitely does.

I don't think this counts as official, but a Mozilla employee replied to a Reddit comment stating that it'll be open source too.

https://www.reddit.com/r/firefox/comments/5wio45/mozilla_acq...

[edit] also nnethercote is a Mozilla employee :)