AI in Book Publishing: A Case Study of Translation and Speech Generation Tools
Mikkola, Pasi (2024)
2024
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2024102226656
https://urn.fi/URN:NBN:fi:amk-2024102226656
Tiivistelmä
The objective of this thesis to was to see whether there are artificial intelligence (AI) tools that could assist with book publishing process, especially, with the stages that require translating or voice and speech generation which could be used in the creation of audio books. The thesis was commissioned by Professional Publishing Finland to whom finding beneficial tools might mean money and time savings. In addition, it is important that the solutions provided by the AI tools are of high quality and usable.
In the theory part the history and the current situation of books and book publishing are introduced. The role of AI in publishing shortly talked as well. In addition, the theory part covers the history of artificial intelligence, the types and structure of it and the models behind the tools used in the thesis. All the tools in used in the thesis are introduced in the theory part of the thesis as well.
Methods in this thesis are qualitive in nature. Four translation tools and three voice generation tools were chosen. All the tools used were free subscriptions which is why not all the functionalities were available. Three texts were translated by each translation tool and one sample text was used when testing the voice generation. All the source texts were taken from a book published by Professional Publishing Finland.
All the results were promising. All four translations tools performed well, and the quality of the translations were quite high. Mistakes were scarce. However, the free versions of the tool work fell for short texts, rephrasing and searching for synonyms but for translating a whole book the paid versions might be better, however, this needs to be researched further before any conclusive can be said.
All the voices of the three voice generation tools were natural but not necessarily suitable for audio books or any other longer texts because they still seem to miss a certain human element which might be disturbing when listening for the voice for hours. However, for shorter texts, the voices could be usable.
In the theory part the history and the current situation of books and book publishing are introduced. The role of AI in publishing shortly talked as well. In addition, the theory part covers the history of artificial intelligence, the types and structure of it and the models behind the tools used in the thesis. All the tools in used in the thesis are introduced in the theory part of the thesis as well.
Methods in this thesis are qualitive in nature. Four translation tools and three voice generation tools were chosen. All the tools used were free subscriptions which is why not all the functionalities were available. Three texts were translated by each translation tool and one sample text was used when testing the voice generation. All the source texts were taken from a book published by Professional Publishing Finland.
All the results were promising. All four translations tools performed well, and the quality of the translations were quite high. Mistakes were scarce. However, the free versions of the tool work fell for short texts, rephrasing and searching for synonyms but for translating a whole book the paid versions might be better, however, this needs to be researched further before any conclusive can be said.
All the voices of the three voice generation tools were natural but not necessarily suitable for audio books or any other longer texts because they still seem to miss a certain human element which might be disturbing when listening for the voice for hours. However, for shorter texts, the voices could be usable.