Overview

Eclosion is a tool which combines speech
recognition, optical character recognition, speech synthesizer and
handwriting recognition combined together to create a new user's
experience like Speech-to-Speech translation, voice command or even
financial management. Speech recognition, OCR and handwriting
recognition can be integrated to the Android keyboard and thus be used
in other applications.

  

Screenshots of the application


Important -5 importants steps to know for making a good optical caractere recognition:
Step 1:be sure the language is correctly set in the option (This option improves precision using a dictionary)

Step 2:the text should be well lighted with clear details. Use the flash if your model has a flash. A picture with low contrast is hardly recognized (shade of gray are difficult to recognized)

Step 3: the text must be aligned correctly (small representations can help you). A bad orientation (few degres of offset or wrong direction) give a poor result

Step 4: adjusts the focus of the image before shooting (let the finger in the shoot button before shooting to adjust the focus)

Step 5: crop the picture to adjust the zone

Help

A separator can differentiate recognition features (top of the separator) and their uses (below the separator). You can resize these elements by dragging the resize icon to the right of the separator. The elements of the bottom (translation, financial management, adding contacts and voice commands) can be rotated by dragging or directly from the menu.

The different features are available from the circle menu. Press "menu" on your device or click on the separator to make it appear.

Optical character recognition

The OCR can use your camera phone to transcribe the text of a photo.
You can choose an image from the SD card or take a photo directly from your device. The OCR recognition is based on a dictionary to refine the results, it is advisable to fill it in the application preferences.
For a good OCR recognition, the text must be aligned correctly (small representations can help you). The handwriting recognition does not work if the text is too steep or in perspective.

The contrast of characters must be important enough for a good recognition, the light source must be large (it does not work in the darkness). The recognition process works better with black writing on white background. The definition must be correct (not fuzzy) and the characters should not be too small.

When the picture is taken, it is possible and even highly recommended to resize the image using the icon that appears at the top right of the image.

Handwritten Recognition

The handwriting recognition mode lets you write handwritten text directly from the screen. The writing of different letters requires a quick learning. The button located on the input area displays the various letters to interpret.

Speech Recognition

Speech recognition can recognize short phrases. You need to have at least an Android Cupcake version and you need to have the Google voice search engine installed to use this service. Speech recognition can be done in English only.

Taking photos

You can take a photo which will be used by the OCR process. The text must be aligned correctly (small representations can help you). Press the button to take photo adjusts the focus of the image.
The captured image is stored on the SD card (and therefore in the gallery) if the relevant option is selected in the preferences.

Gallery

The gallery allows you to select an image to be used by the OCR process.

Translation

The translation mode can translate text from one language to another. The middle icon allows you to simply swap the languages. The translation is done by clicking on the button below right. The two buttons to the right of the text input field allows you to navigate through the history of translations.

Speech synthesis is used for the translation mode only.

By combining speech recognition, translation and speech synthesis, it is possible to do a quick translation without typing any text (Speech-to-Speech translation). A checkbox in preferences allows you to start voice translation by shaking the device.

Speech Synthesis

The speech synthesizer reads the translation text input area. There are two types of speech synthesizer which can be parameterized:
- a web service speech synthesizer. This synthesizer is limited to four languages (French, English, German, Italian) and requires Internet access.
- a embedded speech synthesis that requires prior installation of the TTS service available free on the Market. There a lot of languages supported.

Financial Management

The financial management mode allows to easily manage your finances from a shopping lists or bank statements. Items can be saved in an XML file for use in other software.

Contact mode

The contact mode allows to add new contacts directly into the application from the photograph of a business card.

Voice commands

The speech recognition mode allows to launch different actions (launch a playlist, launch a program , launch a calls and launch of website). The button representing a microphone can record the voice command action to perform. To use voice commands, you should shake the device and activate the feature in the options.

Voice commands can control your phone by programming commands to launch from a sentence or a word. These commands can launch another program, launch a call, launch a music playlist, or launch a website.

Keyboard integration

The various features of recognition (speech recognition, handwritten recognition and OCR) can be integrated into other applications through keyboard. To enable this feature, you must first go to Settings> Locale & text> select "Eclosion".

You now have the opportunity to use these features when selecting a text field in other programs. You can change the features by selecting the arrows above it.
Note that certain restrictions may prevent speech recognition to be fully operational.

The resources are loaded in the first launch of the application and each time you rotate the device. This can be long.

Miscellaneous

Note that the application may consume a lot of resources (CPU and RAM). Cleaning Resources can affect the responsiveness of the application.

You can contact the author of the program at:
eclosion.android(---(at)---)gmail(--point--)com
Please feel free to contact me if you have any questions or comments

my apologize if the following text is not well translated.

Enjoy :-)

FAQ

When the application will be available in the market?

It's available now!

How does OCR work?

Eclosion use an client-server OCR processus and an embed OCR processus, both based on Tesseract. However, only the client-server processus is available for ADC2 submission (you need to be connected to Internet to use it)

I'm not an english user and I can't launch the Speech Recognition. How can I do?

Some androphone doesn't have the speech recognition activated or installed by default. I don't know exactly why. I've post a message in the Google groups here:
http://groups.google.com/group/android-discuss/browse_thread/thread/2a53ec01bdff8e67

If someone know the reason, he could send me a message at eclosion.android(---(at)---)gmail(--point--)com

The voice search APK file is available here This file active the speech recognition in a Magic with Cupcake (Donut seems not supported)

Did you have a support?

No, I've developed alone the application and it take a long time. I don't have any more time now to support the application.

Releases

Version 1.0.4:
ADC2 version
Version 1.1.0:
* Embed or client/server OCR processus
* Better accuracy for OCR: grid capture and image processing
* Faster animation
* Faster image capture
* Memory management improvement
* Optimisation for Donut
* Bug fixes: crash which appears occasionally during the capture is solved