Are you 18 years old or above?
Support my work on Subscribe Star: www.subscribestar.com/dave-cullen
www.computingforever.com KEEP UP ON SOCIAL MEDIA: Gab: gab.ai/DaveCullen Minds.com: www.minds.com/davecullen
Log in to comment
i'm no fan of AI writing myself, it often misses the mark, but with the writers strike and also the actors striking, they are likely going to be replaced by these programs because the show must go on. now i wouldn't mind becoming a writer for something for whatever the going rate is, i have never been payed for my writings before, but i am a creative who likes to create stories.
YouTude and all of it's scripted AI generated video's, with their AI text to speech narration, from it's own video factory, has turned to total shit. It's a cold corpse with a mechanically generated heart beat.
some of them sound pretty damn good though, however i don't mind the comedy shows people create doing the presidents hanging out, the thing is not all of the speech narration is AI generated, i mean we had text to speech tech long before we had AI using it, but here is the destinction, eleven labs voices are just a tts program, now it used AI to figure out the proper tone of the voices to make it sound natural or more human, but it can't do tone very well with proper emotion in the tts dialogue, but it has somewhat been improved to sound out some words better so it doesn't sound like misspelled words being read in the text. it will only improve as time goes on, but then we have the regular tts program that's using the old fashoned computer voice where you just write in the script of what you want and it's generated into a voice file like what this website allows. https://readloud.net/english/a....merican/2-girl-s-voi but eleven labs sounds leagues better in comparison because they used AI, this example was a few months ago now, and it's probably been improved since, but it's a pretty good representation, https://www.youtube.com/watch?v=uaJxHYeqpig
@KEEPER: Yeah been wanting even some reasonably good but basic text to speech software on Linux to convert HUGE 400 page documents into MP3's but sometimes, "Open Source" software is like "Open Sores".... There is a NEW program called Speech Notes - that does text to speech and speech to text (it fucks up Australian English rather badly) and does translations both ways... for MANY languages.. it's impressive, the spoken language is for the most part, excellent - BUT it takes a LOT of processing power OR a long time... https://github.com/mkiol/dsnote There are a few little anomolies in pronounciation here and there but it's basically flawless...... I am going to chase them up on if it is possible to go to a LOWER bits per second processing and speech - as purely monotone robot voices ARE shit... but highly smoothed out human replica voices are very data intensive.... Kind of like basic 3 or 4 step modified sine wave inverters to turn DC battery power into AC for running lights and fridges, Vs the True Sine Wave AC generators, for running computers on... I'd settle for halfway between shit and superb - just on the basis of practicalities.
@Life_N_Times_of_Shane_T_Hanson: that makes sense, especially considering why 11 labs basically makes people pay for it, I mean it's not that expensive, I think someone said it was five bucks a month, and if you're creating a lot of content that makes you money on the side five bucks is nothing. But as for creatives or people who create voice content through a mod like for Skyrim or something or any Bethesda game for that matter, they don't typically make money off of their mod because all mods are free, but they possibly could if they put it in a Collection Pack so that they're not public, but most people don't do that because it limits the public from knowing their work so they end up just asking for donations.
@KEEPER: How I was meaning that Youtube has turned to shit - aside from ALL the really bad crap they have been doing to people... Well that has driven off all the really great content creators, original thinkers and creative people, and so make up for the almost total absence of content creators, they have resorted to running an AI driven video factory, that just churns out a lot of shit - The subject - has like 5 minutes of focussed topic and they just back fill it with 25 minutes of unrelated garbage... The limited run of synthetic voices, and a lack of HUMAN content creators - Mehhhh it get fucking tedious.... AND the people running Youtube DO set the text to speech narration, to speak by adding or omitting "agendas" to the content, to suit their aims... So there is no freedom of choice, only scripted bullshit - "Orange Man Bad" and now "Russian Man Bad" across all of their videos... AND they can program and update the narrated scripts on a daily basis for 10's of thousands of videos by the minute.... One day Tanks of Europe says, "Tanks have long been a tool of war in Europe since their introduction in 1917" - and then the next day, "Tanks have long been used in the Ukraine to defend against Russian aggression" - they just recode the texts to take out some things, add in others, and the text to speech narrators, update seamlessly with the bullshit agenda setting scripts on the fly.... So under the scab of the text to speech engines used by Youtube, that's a problem in it's self, but the scab is the crust covering up what a fucking shithole of deceit, and control and malevolence that Youtube has become.
@Life_N_Times_of_Shane_T_Hanson: "Speech Note" - https://github.com/mkiol/dsnote - has a range of totally EXCELLENT voices, including the basic 1980's robot voice, to really modern voices... Even I am impressed. Now asking them to differenciate between the voices for clarity and data intensity - The basic robot voice uses fuck all processing power, but the REALLY GOOD ONES, use loads of processing - so a 400 page document, being converted into an MP3, while highly legible to listen too, takes ages to convert to an MP3 on anything less than a very powerful and fast processor, with shit loads of RAM etc.. I mean it's very good - but the 400 page conversion takes like 12 hours or more.. BUT in terms of audio quality, it's an excellent MP3 - getting a few mods put in like reading speed, pitch and pause, and a few basic MP3 settings like 44,000, 96,000 as true super high audio quality is a waste of time and file size...... Well some people might want high fidelity like 265 Kbs in stereo...
Browse To Upload