[Roni Bandini] has given the world a new type of music with, BookSound. As the name implies, it takes the written word and turns it into electronic music.
You’re no doubt familiar with audiobooks, which are basically the adult equivalent of having somebody read you a bedtime story. They’re an easy way to boast to your friends that as a matter of fact, you have read that popular new novel when in all likelihood you were probably just half listening to it while you drove to work. But for all the advantages audiobooks have over their traditional pulp-based brethren, they still have one glaring downside: with precious few exceptions, they’re no good to dance to.
BookSound creates electronic music from regular books. Well, electronic sounds at the very least. He says he’s still working on perfecting the word to music mapping, so the results shown in the video after the break are still a bit rough. But even in these early stages there’s no denying this is an exceptionally unique project, and we’re excited to see where it goes from here.
Inside the classy looking 3D printed enclosure is a Raspberry Pi, an OLED display, and the button and switch which make up the extent of the device’s controls. At the end of the arm is a standard Raspberry Pi Camera module, which gives the BookSound a bird’s eye view of the book to be songified.
To turn your favorite book into electronic beats, simply open it up, put it under the gaze of BookSound, and press the button on the front. Because the Raspberry Pi isn’t exactly a powerhouse, it takes about two minutes for it to scan the page, perform optical character recognition (OCR), and then convert that to the beats before you start to hear anything.
If you’re wondering what the secret sauce is to turn words into music, [Roni] isn’t ready to share his source code just yet. But he was able to give us a few high-level explanations of what’s going on inside BookSound. For example, to generate the song’s BPM, the software will count how many words per paragraph are on the page: so a book with shorter paragraphs will consequently have a faster tempo to match the speed at which the author is moving through ideas. Similarly, drum kicks are generated based on the number of syllables in each paragraph. In the future, he’s looking at adding “lyrics” by running commonly used words in the text through a text to speech engine and inserting them into the beat.
We’ve seen practical applications of OCR on the Raspberry Pi in the past and even similar looking book scanning arrangements. But nothing quite like BookSound before, which at this point, is really saying something.