Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: Get conversational practice in over 20 languages by talking to an AI

talk.quazel.com

Post reply on HN
Print view View original

331–340 of 354 posts

Re: Show HN: Get conversational practice in over 20 languages by talking to an AI

#331

Post by LimitedInfo »

treis wrote: post #297
swyx wrote: post #102

am rusty chinese speaker. randomly assigned to converse on Furniture & Locations. --- Quazel: 告诉我吧!你是如何安排你的家具的？ Me (in halting, broken chinese): 窗户旁边有一个沙发，中间是电视和桌子。我的工作就是在这里. Quazel: 所以你的沙发是面向窗户的？那真的很好。我敢打赌，这样你会得到很多自然光。沙发后面是什么呢？ translated Quazel: How do you arrange your furniture? Me: There is a couch next to the window, with a TV and table in the middle. This is where I work. Quazel: So you have your couch facing t…

It's interesting that you got a deeper conversation. Mine was (in spanish): Me: How are you? Q: I am well Me: What do you do? Q: I am an engineer Me: What kind of engineer? Q: I am a civil engineer That's when I got the pay/subscription wall.

it's free right now, you just need an account

Re: Show HN: Get conversational practice in over 20 languages by talking to an AI

#332

Post by LimitedInfo »

fpierfed wrote: post #301

I tried in Italian. Speech to text is impressive: kudos. The conversation was definitely weird and very repetitive. It needs more work I think. In any case, great idea and I am sure that you will soon enough get a fantastic product out.

sei italiano? pensato che la conversazione era buona ma io ho solo studiato l'italiano da 6 mesi.

Re: Show HN: Get conversational practice in over 20 languages by talking to an AI

#333

Post by xhevahir »

tomtomtom777 wrote: post #312
aix1 wrote: post #303

Earlier quoted context omitted.

> Believe me, if I the syntax in Dutch would keep, get you a very strange result. (This is a literal translation of a correct Dutch sentence.) Of course there are differences, but both Dutch and English are West Germanic languages (descended from proto-Germanic) whereas Portuguese is a Romance language (descended from Latin). Here is a handy tree showing how various languages are related to each other: https://d1sjtl…

Such a tree mostly describes word similarity. But translating words is not what an AI (or more generally a computer) has problems with. It's word ordering and idioms where AIs stand or fall. Things that show very little correlation in the evolution of language.

> Such a tree mostly describes word similarity.

It describes genetic descent.

Re: Show HN: Get conversational practice in over 20 languages by talking to an AI

#334

Post by holmesch »

Hadjimina wrote: post #183
phpisatrash wrote: post #2

It really looks cool. But, why it doesn't have English? Or meybe if it does have, why isn't showed first? People who use language learning apps are focused in learning English most of the time. Still those who speak fluently are trying to improve their speaking and listening. And the first language which appeared for me was Brazilian Portuguese which is my mother language. [Edit] I was trying the Brazilian Portuguese…

Hahah I have a super silly response for this one: It's simply because we have not gotten around to translate the UI and everything into other languages besides english. That's why we simple assume people know english, since they need to be able to know at least some english to use it. But it's high on our to-do list.

You don't need to translate English UI for training English because there are quite a lot of person don't talk / listen well even though he can write / read English quite okay.

It's because Reading / Writing can train oneself but conversation must be training in both sides, which always inadequate in some area, such as HK, since we don't talk in English but we can read quite well

Re: Show HN: Get conversational practice in over 20 languages by talking to an AI

#335

Post by tidenly »

Ran into a problem immediately in Japanese.

I asked お名前は何と言いますか？(Whats your name?) and it correctly said Sarah, but then when I asked 何人（なに人）ですか？(Where are you from?) the transcription into kanji caused the なにじん(nanijin) to be interpreted as the most common way to read those kanji なんにん(nanninn). So she responded to the question "How many people are there?" with "There are 6 people in my office".

I don't know how easily fixable this is. Either context needs to be taken into account, or the original phonetic transcription should be kept underneath the kanji.

Re: Show HN: Get conversational practice in over 20 languages by talking to an AI

#336

Post by amatecha »

Cool idea!! I'm just trying it now with French. It asked me where I want to travel, so I said Paris, then I said I love Lyon too. Then it said "have you had a chance to visit?" ... well, duh. I tried to write a response in English and then translate that to French (because I don't know how to say "yes, that's why I said I love it" in French) but that doesn't seem to work, and the app is kinda frozen up now. I can see a console log twice "send w.o. punctuation" , but I'm not sure if that's related (I did try to translate my English to French twice)...

BTW I see a ton of network errors trying to hit Mixpanel. I hope your code is written such that Mixpanel calls aren't blocking calls, because tracker-blocking behaviour in Firefox or other browsers will stop those services *such as Mixpanel or GA) from loading.

Re: Show HN: Get conversational practice in over 20 languages by talking to an AI

#338

Post by forgingahead »

englishrookie wrote: post #272
melvinmelih wrote: post #223

Earlier quoted context omitted.

Dutch is a sister language of English so I’m not surprised

Believe me, if I the syntax in Dutch would keep, get you a very strange result. (This is a literal translation of a correct Dutch sentence.)

Yoda was Dutch?

Re: Show HN: Get conversational practice in over 20 languages by talking to an AI

#339

Post by jaclaz »

bobajeff wrote: post #47

In my opinion the issue with most language programs is that they focus too early on speaking when listening should be the focus early on. Really speaking shouldn't even come into the picture until exposure of about a year and something like 15,000 hours of good comprehensible input.

365x24=8,760 hours in a year, 15,000 is nearly 2 years 24/7, maybe you meant 1,500 that is roughly 2 years, 2 hours per day? 2x2x365=1,460? (or 1 year, 4 hours/day)
Post reply on HN