Record voice online

Until now, you had to have an existing audio file (recorded with your favourite recording software). This step is just made a lot easier with the newest version of the application. Now you'll be able to choose if to

1. Upload an existing file or
2. Recording the voice directly on the site (without external applications)

Just click "Record" and start talking. When finished, click "Listen" to hear your recording. If you are not happy with the result you can record it again. When ready, click "Upload", and that's it.

Unfortunately, the javascript recording API isn't available on all browser, but if you use the latest Chrome or Firefox -- you should be fine.

PS. Next thing in my tasks pipeline -- make the design better.

Exercise editor

Just deployed a new version. The biggest change is the way how the text is synchronized with the audio. Basically it now happens in two steps. In the first one you must enter the text in the following format:

Ideja {The idea} ovog sajta {of this site} je {is} da ljudi {for people to} objavljuju {publish} kratke snimke {short recordings} govora {of spech} ....

Basically you write your text in the original language, and then the translation in curly braces.

In the second step you must "synchronise" your text with the audio. Just click on the chunk of the text and then you need to increase or decrease the time, until you get it right.

It sounds more complicated than it is.

Dictation

It is one thing to just listen (and to try to understand what the speaker said). It's something completely different to listen actively. Today upgrade contains a active listening quiz. It's called dictation.

You are asked to listen closely and then type what you heard (in your target language). As you proceed, the voice will speak the new portion of the recording. If your text is correct -- it will be shown in green, otherwise it's in red. If you are not sure about the next character -- just type ? (question mark).

First example

Just finished a first "proof of concept" of the idea.

An example audio+text+translation is here: Ideja ovog sajta ("The idea of this site" in Croatian).

The idea

The idea of this project is simple. It will be an application where users who speak a language will upload short voice recordings along with the text and translation. Text and translation will be "synchronised" with the audio. Users will be able to listen to the audio with selected the audio the speaker is currently speaking, and same for translation. Translation will not be word-by-word but for chunks of logically connected words (phrases, names, article+noun+adjective, ...).

With that information I'll create additional quizzes for every recording.

Users who upload their recordings don't need to be professional speakers. Users who will use the site regularly will be asked to contribute with recordings in their own language.

Before the start

I'm very bad with languages. I speak three (Croatian, Italian, English) and a half (Slovenian). When I say I'm bad at languages I mean that I'm bad at consciously learning languages. I learned Croatian (my native language), Italian and Slovenian by just being part of a culture where those languages are spoken.

But English... That one was hard. It took me 20years (and I'm still far from perfect).

I decided to try to learn a completely new language... Found some grammar books, started adding words and translations in ankidroid. Then, I turned on the radio, found a random station in my target language and... What I heard was noise. Not that this language sounds like noise, but it was informational noise. Just sounds not carrying any information. Not only I was unable to understand a single word, but I wasn't even able to recognise where words and sentences start and end.

When I tried to listen to speakers in (acted) recordings in various courses -- everything sounded a lot clearer and easier.

That's how I decided to create a web application which I'll use in parallel to my language learning.