Bhashini AI Solutions
Connecting People Through Language Technologies
ಭಾಷಿಣಿ ಎ.ಐ. ಸಲ್ಯೂಷನ್ಸ್ - ಕಿರು ಪರಿಚಯ
ಭಾಷಿಣಿ ಸಂಸ್ಥೆಯ ಶ್ರಮಕ್ಕೆ ಸಿಕ್ಕ ಪ್ರೀತಿಯ ಗೌರವ
Who we are and what we are up to?
Digitizing Indic Literature
Digitization of printed books into eBooks and audiobooks
We undertake digitization of printed books into eBooks and audiobooks for making them available on modern digital platforms such as Amazon Kindle, Apple Books and Google Play Books.
We leverage the high accuracy, award-winning OCRs and TTS developed at IISc and IISc startup, for speeding up the creation of eBooks and audiobooks, and make them available at a fraction of the current costs incurred for manual typing/recording.
The copyright for the newly created eBooks and audiobooks remains with the original authors. Money generated from online sales of eBooks and audiobooks goes directly to original authors, thereby creating additional source of revenue for them.
Benefits to readers: listen to audiobooks during daily commute, accessible for visually challenged people (DAISY books), reach Indic speaking population in other countries (US, UK, Canada etc).
Creation of eBooks and audiobooks - our workflow
Collect physical book, remove binding and scan each page of the book.
Upload scanned images to server
Mark text and image regions and validate skew correction.
Run high accuracy, award winning OCR from IISc.
Proof read & manually correct the OCRed text, especially the words marked by OCR as having low recognition scores.
Mark paragraph ends and section/chapter links.
Generate EPUB eBook.
Generate audiobook using high quality, award winning TTS from IISc startup.
Upload eBook and audiobook to Google Play Books.
Digitization of libraries and newspaper/magazine archives
Preserve historical archives through digitization as well as make them searchable.
Options to save digitized content in international standards such as METS-ALTO XML, searchable PDF (text behind image), EPUB, .doc etc
Creation of large volume annotated datasets for language technology
We also undertake creation of large volume annotated datasets needed for language technologies such as Optical Character Recognition (OCR), Text to Speech (TTS) Synthesis, Automatic Speech Recognition (ASR), Online Hand Writing Recognition (OHWR) and Machine translation. Some of the successful datasets created are:
Speech recordings needed for Kannada ASR (Automatic Speech Recognition) project. Speech from 500 unique speakers in Bangalore rural region aggregating to a total of 250+ hours of speech was recorded for the needs of Kannada ASR project at MILE lab, IISc.
Datasets for testing OCR (Optical Character Recognition) accuracy for Konkani, Tulu and Sanskrit documents printed using Kannada script. https://github.com/MILE-IISc.