黑料网

News

Audiobooks can today be created with AI

The first audiobooks were created in 1932 on vinyl records. Noticeable audiobook market growth started later in the 1960s by the advance of technology and cassette tapes and in the 1980s by compact discs. Today audiobooks can be created with artificial intelligence text to speech tools and applications.
The Fifth Wave - BRIE-ETLA Collection of Articles book cover

The Department of Industrial Engineering and Management (TUTA) has published its first AI audiobook (four chapters of the book) conducted as an auto-ethnographic study project during the summer 2023 from the book 鈥楾he Fifth Wave 鈥 BRIE-ETLA Collection of Articles麓 edited by Timo Sepp盲l盲, Tomasz Mucha and Juri Mattila. Autoethnography is an autobiographical way of conducting research, in which the researcher's own experiences and the field notes written or otherwise produced from them form the central material of the research.

The motivation behind the AI-based audiobook project was quite straightforward. A colleague of senior lecturer Timo Sepp盲l盲 from the machinery industry once mentioned that he didn't have the time to read Timo鈥檚 book, but he could listen to it while commuting to work or returning home. 

How was the AI-based audiobook done? 

鈥淚 read 20 and my colleague Tomasz 50 sentences to the AI tool. We then discovered that 50 sentences produced much better sound quality, so I went back to the studio and read another 60 sentences. Based on these, the AI tool cloned our voices and used those as narrator for the book鈥, explains Timo.  

The producer of the book, Laura Loikas, is a student of industrial engineering and management at Aalto. She was hired for a month during the summer to work on the project. Laura compared and chose the technology to be used, did the cloning of the voices and edited the articles of the book so that AI could read them. Then she produced the mp3 files and edited the parts where AI made mistakes. Altogether the production costs were around 200 Euros without salary costs. 

鈥淓ducational materials are used to be produced in written form, but we see that in the future any written teaching material can be produced in audio by using AI tools and applications. This will also give new tools to the student to learn things regardless of location. As a result of this project, we can now transform written materials into audio with reasonable quality, reduced costs, and increased efficiency. Next time, we will provide instructions for this process鈥, says Timo. 

  • Updated:
  • Published:
Share
URL copied!

Read more news

Lecture hall with students watching a man present slides on electrically heated textiles at the front
Research & Art Published:

Babak defends PhD thesis on electrically heating bio-based textiles

Babak Abdi publicly defended his doctoral thesis.
Lecturer in a modern auditorium presenting a slide titled 鈥淢erging Textiles and Electronics 鈥 and Beyond鈥.
Research & Art Published:

Invited Talk: Swedish Expert Explores the Future of E-Textiles

The Textile Chemistry Group of the Department of Bioproducts and Biosystems organized an invited talk on March 26.
Four men are smiling while playing table soccer game.
Cooperation Published:

AI enhances security screening 鈥 Master鈥檚 thesis improves See Through Solutions鈥 camera technology

New AI solutions are being produced in FAIR ecosystem. See Through Solutions, a deep-tech startup, gained momentum for its development work from a master's thesis.
Modern exhibition room with mannequins in green and white clothes, bowls and a beige bathtub on white plinths
Research & Art Published:

Finnish wood-based innovations on display in London

The exhibition showcases Finnish bioeconomy solutions for both industry and everyday consumer life. Origami-based FOLD materials offer more sustainable solutions not only for packaging but also for many other applications.