eventually I want the site to be able to stand alone on a "Furby" style device, as well as people being able to install an app which people can say things, or type things into and be able to get endless audio or text based around the thing that they said to it, for instance if I said, "tell me a story about Jesus, but base it in the dystopian future where we all live on Jupiter", it will find all the things its heard about Jesus & Jupiter & the dystopian future & using Zipfs law of words & spoken words it has heard before, begin endlessly speaking about the thing it was told to speak about until its told to stop or change the subject, it is likely to say something about the moons of Jupiter being more habitable, if it has been listening for a while... the idea is to let it listen to everything and then turn that into text & index the text, so if I ask it to tell me about jesus & jupiter for instance, it will index that question also & the next time I ask it about Jupiter it will have a higher chance to bring up Jesus & the dystopian future...
Thats a long way away though but the more it listens to and indexes the smarter it will get & the more accurate the Zipfs law of words list will be.
OYMYO began as a project which aims to make "Deaf Radio", I use the site to index transcripts of conversations from around the world, I have accounts which people can invite to their conversations on the application "Zello", the accounts then begin listening to the spoken words and turning them into text, which is what the site is full of, each day we are getting about 30 days worth of audio which is added to the transcription que & if we dont transcribe it, it is likely to never be transcribed & maybe even not heard in many cases. It transcribes a minimum of 24 hours worth of spoken word per day but depending on how many computers we have running we usually do around 3 days worth of audio a day, the aim is to get the que empty eventually but we are relying on free apps mostly & virtual audio cables can be a nightmare to get set up... the reason the text is often unreadable is because the speech to text software is still learning & the pronunciation of spoken words by people are often interpreted as different words due to it not having the same way to understand accents as a human, it also has to translate foreign languages too so the text seems even more random & abstract, eventually I plan on using something like Grammerly to improve the replies.
I also try to find people to join the project & take part in the conversations which are going on in Zello, people can add the user "www.OYMYO.com" to their channels & then we can get a quick idea about the content of the channel before inviting new users to that channel, people use it also to "monitor" channels where human moderation is lacked, which is why there seems to be a lot of profanity. OYMYO then tries to encourage the rookie users to help the efforts of the "fallen" channels, to stop bullying etc...
sorry about the long reply, its a good excuse for me to brainstorm the project and its long term plans.. it began as a side project (I made a quite successful clothing brand in the early 2000's) which hosts parties around the world & gained a few followers due to the music, we now use that brand to avoid having an advertising based model for OYMYO but in the long term we want to gain substantial funding, probably from government, to deliver the aim of developing the free "ultimate hearing aid". we have already began putting a team together to mod the application zello, then i am approaching government for a 10k grant to buy the licence of the zello app, which will be modded to include speech to text for deaf people, so it will work like a "Walkie-Talkie" for hearing impared people, dokuwiki will hopefully be used to index the conversations within the modded zello app, which may be an advertising based model in the future, with profits split among the contributors.
I have been putting together custom books, which I send to key people to advertise the efforts of OYMYO. For instance, key speakers, celebrities, politicians & coders, anybody who can get the app delivered and also to grow the network already in place.
it is useful more and more as time goes on & the more information it transcribes the smarter it should get & the easier it is to spot new things or irregularities, the Zipfs law of spoken words which it has produced is probably the biggest in the world freely available & down the line (maybe 20 years) will be one of the main aspects of the project, which in my mind will be like a "Furby" but installed as an application, or as a standalone device.
So eventually the end user will be able to get a dilution of new information it has found based on what it predicted was going to be heard in previous patterns of speech and what was actually heard, for example, if "earthquakes" are suddenly being spoke about more regularly than "earthquakes" position in the Zipfs law of words, it will push this up as a irregularity like an automatic news feed.
thanks