about audiobooks
I have delisted most audiobooks for the time being. The quality of AI-generated voices based on real human speakers is not yet quite up to the quality standards I would like.
Still I stick to my comments below about preferring—once they are more perfected—neutral readers, as those provided by having the audio generated by computers, over just about all actual human voices. I’m leaving this comments on this page, just for the record.
Still, unless the technology is perfected, I’ll stick to either reading the books myself—not likely, as that would seriously eat into the time and energy I need to invest in actually completing several unfinished novels—or just skipping audio editions for the time being.
————————
Leftovers of the older version of this page…
📝 I recently downloaded an audio book, rendered by a major and very reputable ‘real person’ reader, of Ben Bova’s Voyagers. Don’t know if it’s just me, but I found that the reader’s voice interfered with the story so much that I stopped listening and decided to reread the novel in hardcopy instead. All the Ben Bova audiobooks are read by the same person, which means I’m not going to listen to the sequels either. I found similar issues with other ‘real’ human voices when sampling audiobooks from our local library through BorrowBox. And then I tried a snipped from an Audible reading of Jack Vance’s The Green Pearl. Here I found that the reading affected me very differently; the voice, enunciation, emphasis, timing… Maybe it’s just me, a deep admirer of Jack’s work. But I suspect that this reader definitely would not have been Jack’s choice. Not sure which of these two ‘human’ readers were more objectionable. Again, I hasten to add, to me at least.
By contrast, as an example, the digital (US) ‘Warren’ voice I’m using for the Tethys novels is far less intrusive on the stories. Could be because it’s more neutral, yet capable of expressing emotions and even varying between reflecting male and female speech. Definitely not perfect, mainly because of the AI guiding the voice and its cadences, intonations, pronunciations etc. Still, the flaws only minimally interfered with the story. To quote Jack Vance: “A reader is not supposed to be aware that someone's written the story. He's supposed to be completely immersed, submerged in the environment.” Replace ‘reader’ by ‘listener’ and ‘someone’s written the story’ by ‘someone’s reading the story’…
A lot of readers imprint too much of themselves—e.g. through emphasis, tone, pitch, timing, as well as those other elements a live storyteller might use at, say, a book reading or around a camp fire or wherever s/he faces a live audience— and thereby also their interpretation of the story; what matters and what doesn’t; what the listener should pay special attention to, even—possibly subtly though implicitly—their like or dislike of elements or parts of the story and its characters. That doesn’t happen in a physical book during its reading, be that visual, e.g. hard copy or screen, or by touch, such as through a reading with braille. There are just the written words connecting author and the reader.
Audiobooks, at least in my opinion, should preserve, as much as this is possible, that kind of connection. One of the reasons why I think I may actually prefer digital voice; the version derived from a real human voice, but which does not interpret, but renders as faithfully as possible, sticks as closely to the author’s text—words, sentence structure, punctuation marks’ significance, etc— as possible, allowing the listener to make up her own mind and construct her own understanding of and relationship with the story. That needs an unbiased human reader. Sounds like an irresolvable contradiction, but it looks to me like advanced digital voices reading a text are getting closer. Maybe currently improving audiobook technology is one of the better aspects of AI used in a beneficial way. Time will tell..