JZL003

Also, not as granular, but instead of the + suffix, add a dot in a weird place. So

n.ame@gmail.com or nam.e@gmail.com . Many SMTP servers respect periods as differentiating emails, so services can't delete them. It doesn't help you stop spam, but you can add a gmail filter that n.ame@gmail.com is put in a separate label. And it's very fast to type, easy for non tech-y people

I guess it's harder (although not impossble) to send email from this throwaway address, but that has never come up for me, for external accounts

There are lots of ways to do this. Postfix is nice but a little heavy. The simplest and most functional way I've found is https://github.com/0xERR0R/mailcatcher since all it does is forward the emails. You can even use a throwaway gmail SMTP so it doesn't get send to spam

Easy to set up on a rpi/cheap VPS, as long as you have a hostname. And while you're there, look for a short domain name so it's fast to type (on credit card kiosks). You can get cheap short non-standard TLD's like .li. I got a 3 character domain for $5 a year, as short as bit.ly, but just for me

I tried a couple of different programs for personal use and udt/wft wouldn't compile for me even after 2 hours+ messing with it

CERN's fdt https://github.com/fast-data-transfer/fdt was way better. I did use it over ssh port forwarding which artificially slowed things down, but it saturated my uplink to 300-400 MB/s

I've grown to really rely on R pipes, it's just how I like thinking about problems with small short lambdas. I wish python had better support (and better lambdas, without typing out 'lambda' and with tuple arguments) but for when I need to use it, toolz's pipe is pretty ok as long as you use lazy map/filter/etc

It's just so close to being as good, but not quite

I like YT-DL but I'm amazed it still works. Google pushes use of widevine for netflix/etc to try to around this issue. I know there's a difference b/w 'premium' netflix content where it's only professionals vs youtube where it's semi-ad-supported, but they could just pull the widevine switch on principle if they got serious

It's long but really good, I highly recommend Folding Ideas's video about NFT's (and bitcoin) https://www.youtube.com/watch?v=YQ_xWvX1n9g . He says it much better than I could the often vacuous greed and bad faith arguments of it

I saw the same thing. I watched the video and they seemed to choose the 4 primary pigments to be close to CMYK, but I was also surprised they didn't make a lookup table for other CMYK pigment alternatives. Seemed like it wouldn't be crazy to do and gives another nice artistic knob to turn. And if it didn't affect much, that would also be convincing that they weren't too sensitive to the 4 pigment selection which is also nice

(Although, academic paper cherry picking aside, I was pleasantly surprised how visually close some of their examples were to the read world, for different pigments)

I'm pretty sure it is using pure DNS recursion, so if you ask for abc.google.com, the pi itself will ask the com TLD DNS server for google.com's DNS server, then go and ask google what abc.google.com's IP address is

So you're not asking the ISP for abc.google.com, but because they provide the internet, the ISP will see all your requests, including the UDP request to the .com TLD, google.com's DNS, etc

Yes and it's actually a sensitive test for fit. If you get fog then air is leaking out unfiltered, so when you breathe in, it's also not filtering that part.

It's imperfect bc you can have gaps around your chin which won't fog up. But it's one of the nice parts of the good filtered masks. And in the winter it helps keep the bite of cold away

On older kindles at least, they had time not date. If you keep it off wifi(so manually copy the books from the manage Kindle page), your local copy never expires. It still returns for the next person but you can read it later

On my phone so can't respond fully but if you select the <video> element in chrome and Firefox, you can control the .playbackRate attribute. Extensions can be useful but also abused, this is simple enough to do with a bookmarklet or manually

my problem with non-fiction TTS (or at least the non-fiction I read) is it often involves diagrams or figures and that really doesn't work with TTS, sadly. But histories are nice

For a technical audience, I would look at some TTS (text to speech) programs especially by google and IBM.

It's definitely robotic and nothing like a nice narrator but audiobooks are amazing regardless. And your mind actually starts filling in the emotional blanks. It can also be really cool to use like internet archive's scanned book's OCR -> TTS and make an audiobook from a cool old book that would never be professionally narrated

And for anyone who listens to a lot of audio, I'd look into using an audio equalizer. Pulling down the high frequencies (especially for some woman narrators) makes it more comfortable after many hours of listening. On android the "Smart audiobook" app has this and it's really nice. Maybe some headphones/android phones can do this globally

That's so random, I didn't know he wrote annoy or that it's under spotify's aegis. I use it so many times a day as part of UMAP in R for looking at single cell data. It's always funny when it talks about writing the annoy index, b/c sometimes I'm feeling annoyed myself. Instant RSS subscribt

Thanks for writing this and annoy!

Cool yeah that makes sense, makes total sense for ML where you just need to run over epochs, less clear for other workloads.

After looking around I thinking more about CRIU/docker suspend. The google stars aligned and I found this https://github.com/checkpoint-restore/criu-image-streamer + https://linuxplumbersconf.org/event/7/contributions/641/atta... which actually seems perfect. I wonder how fast it is

Edit: Also no GPU support AFAIK but https://github.com/twosigma/fastfreeze looks really nice, turnkey. I wonder if I write to a fast persistent disk if I can get higher maximum ram than over the NW

(or, hacking on a checkpoint idea, have a daemon periodically 'checkpoint' other programs so even if it's too slow over 60 seconds, revert to the last checkpoint. Even an rsync like application where only send the changes)

Oh I didn't see much in Nimbo with a quick glance but reading more closely > We immediately resume training after interruptions, using the last model checkpoint via persistent EBS volume.

Makes sense, just save checkpoints to disk. What I'm doing is more CPU bound and not straight ML so less easily check-pointed, sadly. Cool though, it's worth jumping through hoops for 70% reduction

I'm interested in how they 'hibernate'/save the state of the instances within the shutdown time limit. I was also looking into this for myself, there are ways of using docker to save the in-process memory a-la hibernate, which would work well with this. But, especially for GCP where you only get ~60 seconds between the shutdown signal and hard stop, I was worried that it wouldn't save it fast enough. I often work on pretty high-ram instances and thought even saving from ram to disk would take too long for 150-300GB ram uses.

I hadn't heard of nimbo, maybe I can read how they're doing it since it's open sources. Does anyone have any idea how they're saving state so fast (NVME SSD disk?)

I personally like ibm tts slightly more than Google although it is more expensive. I think Google, but definitely other startups also allow you to train the model to be closer to your voice. Since you might have many recordings of yourself before this, could be cool. Comment if you can't find it and I can look

It's none of my business how long term it is, but if it is long term look into stenography, especially open source opensteno and plover. Old talk but https://youtu.be/Wpv-Qb-dB6g

It's what court reporters use to transcribe at the speed of text but for you it could mean giving text into a tts software at the speed of normal talking. I've only tried it briefly but if you take the time to learn it, you can comfortably get to normal speech speeds with no hang fatigue. It's one of the reasons opensteno is so nice bc it also has cheap steno keyboards which are normally thousands and locked behind industry. Comment if you want more links! I can add when I'm by my computer

Also just bc the link is buried, see https://doctorow.medium.com/i-quit-9ae7b6010c99

He traded smoking for buying a new laptop every year. Now that it's been years, I guess he could quit and not buy a new laptop. But also people do more wasteful things. I do understand though, I drive laptops into the ground over many years but still 4-5 years per laptop

Yeah https://cloud.ibm.com/catalog/services/text-to-speech it's so smooth

That is pretty nice, one of the best collection of voices I've seen and the best interface

Google gives 1 million characters per month free which I don't often go over, but this will be really useful for when I do

I don't want to be unappreciative, it's amazing that this is possible much less free, but when you spend hours listening to it every day, the cracking and warbling do get old. I think there are better models I've heard snippets of but the truly amazing thing about google's is how robust it is to very weird words

(When I tried all the public cloud offering's, IBM's was the marginally nicest AFAICT but it was the most expensive with least free quota)

Anyone have a nice opensource/pretrained TTS model they like using? I use google's wavenet TTS heavily to create 'audiobooks' (especially from archive.org which is great for old books). But it's pretty expensive

I periodically look for new versions and, while the examples sound better, they fall down really hard on other text.

Seems pretty similar to a couple of things, not least terra (although that is nominally more biology focused). Where terra allows setting up pipelines for data as well

My work uses GCP not AWS so I've been experimenting with google cloud run (it's actually parallelizing R code so need the docker container infra). My only problem is that I have very bursty useage and the auto-scaling is too slow. I made one attempt [1] to encourage larger allocation but don't know another way. Do people have experience with this

[1] Slightly costly but ~ 5minutes before I need it, I set the minimum instance size to a larger number so it starts ramping up, then when I'm done I lower it

Can you say more, I'm curious.

Is it automated in some way during web browsing, remembering to copy to a folder when you enjoyed it enough, or do you use a reading app/e-reader to read them so they're already downloaded

hah! yeah that's what I was looking for too

Especially for US people don't forget overdrive! It's through your public library and often has a great selection

I really like google text to speech and use it for my own custom audiobooks, I've tried google's microsoft's, IBM's, and a few other research ones. IBM's sounds slightly better but has a much more restricted free monthly tier, google's and microsoft's has 1 million free characters per month which goes pretty far.

Like others are saying, it's slightly robotic but I've started to listen a ton by TTS and you definitely get used to it (you even start to hear inflection in it, which is cool). I use android smart audiobook app and you can control the sound levels, turning down the high pitch aspects also helps to make it easier to listen to for longer periods of time

For HN folks, there are some pretty reasonable research projects, especially by nvidia (glownet) which you can run yourself. They sound relatively similar but the training voices are much more restricted and not as good. If anyone knows of a github/etc with a nicer diy TTS I'd be interested. The best I've seen which is customizable is https://github.com/espnet/espnet but I had trouble getting it to work, then getting it to sound ok

(For anyone else going DIY I'll warn you that the failure modes for TTS is some unerring frankly creepy sounds. Google's TTS fails very well, even for strange words, and when it gets very confused it spells it out. Some of the research ones go into haunting unrelated syllables, sometimes repeating for 10s of seconds

I honestly don't see that on any reputable retailer I could find