Hacker News Reader

Comment by csande17 | Hacker News Reader

The "Helena" sample contains a pretty good test case of this system's ability to guess where emphasis and pauses go:

> I know Bill carried within him deep currents of spiritual yearning that he found easiest to express through the beauty he saw in all places wild.

(I checked in an online sample of the ebook: there is no punctuation in this sentence.)

Unfortunately the AI completely faceplants, placing an enormous pause right in the middle of the phrase "all places wild". It actually changes the meaning of the text, making it sound more like "...the beauty he saw in all places. Wild!"

I wonder if any of these AI speech synthesis tools come with an editing tool that you could use to tell it not to put the pause there.

View on HN

I guess this is more of a case of garbage in - garbage out. The original sentence itself is not well-structured. But people who write like that and don't edit it; won't care of how the AI reads it.

I don't feel like this is a product for carefully producing audiobooks, but to create them by the pound, so to speak. I'd say it's a move for the "make your own business through audiobooks" people [1] -- very strange for Apple.

[1]: I didn't know this audience existed until I saw this video on it (and the cons that happen) from Dan Olson: https://www.youtube.com/watch?v=biYciU1uiUw

I can understand that sentence fine, I don't see anything wrong with it.

Writer here, it's a terrible sentence.

"I know Bill carried within him deep currents of spiritual yearning that he found easiest to express through the beauty he saw in all places wild."

"Deep currents of spiritual yearning" is both cliched and unspecific / unclear.

"That" is superfluous.

It's unclear how Bill is expressing the beauty he sees, and the sentence structure implies he's somehow responsible for the natural beauty.

"All places wild" is wilfully awkward and anachronistic.

There are countless better ways to say the same thing. For example the tone would be similar and the sentence more concise just to say:

'When Bill spoke of the beauty of nature, I could sense it inspired spiritual feelings in him.'

Or simply: Natural beauty inspired in Bill a yearning for connection to something spiritual.

Neither are great - because the central thought is unclear. Writing is to a large extent the process of expressing a thought or feeling. Clarifying exactly what one wants to say is a central part of writing and editing. The author seems to have failed to clearly define their idea or emotion, so its expression is decorated rather than clarified however you phrase it.

> "That" is superfluous.

I always use the trick when I write of reading the sentence without the "that"...still makes sense then you don't need "that". Mostly.

> Writing is to a large extent the process of expressing a thought or feeling

Neither of which we believe a machine to be capable of conceiving.

It doesn’t matter if the sentence is terrible writing. It can still be easily read out loud correctly, but not by this computer program.

Could you do Hemingway, McCarthy and Joyce next? It's untenable they wrote 'unspecific / unclear / superfluous / awkward / anachronistic' language such that our robots cannot render to speech.

They write literary fiction, though. The selected passage is listed under "Nonfiction/Self-development"

"The Distant Shore: Stories of Love and Faith in the Afterlife"

It's a book of stories about her dead husband. Non-fiction, yes, but it's not exactly "How to Lose 10 Pounds in 10 Days".

Mind if I ask where you're from? While not really adding anything or altering the meaning (so technically it could be called superfluous), removing it here to my British brain makes the sentence sound lazy and I was definitely taught in school that it should be there. I do also agree that the sentence is terribly written in general though.

Sure. I'm Irish. I've been a professional writer for about thirteen years. Formality of sentence construction is context dependent. In prose there's usually little purpose and no solid grammatical rule for retaining superfluous words. Depending on the pace and rhythm of the piece - which in turn dictate interesting things like the readers perception of time. For example you can make time flow faster by using brief, truncated, staccato sentences. Or stretch it out with more formal, grandiloquent sentences. You can also convey the informality of a relationship, or even the physical structure of space or an object in much the same way. These are elements of voice - the character or tone of the sentence or paragraph, which is contextualised by the overall piece. I'm not articulating theory here - more trying to convey how I and other writers intuitively learn to play with prose.

Fantastic answer, thank you. Thinking about it, it's literally decades since I last wrote creative prose, maybe my formal writing muscle memory was just kicking in here.

Reader here, it's a perfectly fine sentence.

Thanks for this. Can you do Go Dog Go next?

You might be right with some of these criticisms about the ideas and style of the sentence, although I think most of these come down to taste and context.

Grammatically, the sentence is easy to parse and a native English reader would understand how to say it out load.

Native English speaker here: I flubbed the first time I read it - specifically the 'all places wild' bit at the end.

I don't find anything terrible about it. I'd probably break it into two sentences but that's mostly down to my style.

> Writing is to a large extent the process of expressing a thought or feeling.

Agree.

> Clarifying exactly what one wants to say is a central part of writing and editing.

In professional/technical writing, sure? Authors have been known to have different writing styles, and the style may be ambiguous or convoluted to intensify the thoughts or feelings the writer is trying to convey.

PS everyone posting here is a “writer”

Context for the sentence: https://books.google.com/books?id=-v4GEAAAQBAJ&pg=PT20&lpg=P...

> Linda Hale Bucklin made a pact with her husband, Bill, to communicate after death. Here she shares her personal experience and other stories of love and faith in the afterlife.

Bill died after 40 years of marriage.

The rest of the paragraph:

> Bill loved and found solace in nature. Often, we voiced our wonder of the sight of a full moon peeking above the hills of Stinson, rising heavily until it pushed free of the horizon, a perfect circle in the dark sky. On nights with a new moon, we would walk to the end of the beach to find our favorite constellation, the Pleiades. I know Bill carried within him deep currents of spiritual yearning that he found easiest to express through the beauty he saw in all places wild.

I also thought it was a bad sentence, but thank you for giving me solid reasons why that is so.

To me this feels like the automated video slideshows that Apple Photos (and undoubtedly Google Photos) makes for you. Perfectly fine, but indeed even on a casual watch you notice mistakes/imperfections you'd never make if you were producing such slideshow manually.

But that's the thing...is "perfection" worth 3 hours of video editing for something you casually consume?

I think almost any audiobook listener will vastly prefer a serviceable but imperfect audiobook when compared with no audiobook at all.

Agreed. I bought an old Kindle for traveling because I could use its TTS to listen to ebooks. Now that's a common feature of most reader apps, yet at the time it was rare and even newer Kindles had dropped TTS.

I find those slideshows unintentionally hilarious - ten photos of my kid interspersed with a flash photo of the back of the washing machine and some cable that's hanging down underneath my car, an accidental screen grab of a text message, all treated with the same importance and jaunty library soundtrack.

This made me LOL, as I’ve had this happen many times. But much less so in the past 2 years, around the time iOS Photos started integrating with Apple Music.

Actually now you come to mention it I haven't noticed it very recently - maybe I've stopped looking at those suggested slideshows, maybe I've turned a setting off without realising it.

> But that's the thing...is "perfection" worth 3 hours of video editing for something you casually consume?

Any book of any length took countless hours to write and edit. Yes, I think it's worth a bit of extra time for a human to go through and read the thing aloud.

If the alternative really is no audiobook at all... okay, I guess something is better than nothing. But on the whole, I'd like publishers to just record more audiobooks, and I'm concerned this technology will result in fewer "real" audiobooks being produced.

>But people who write like that and don't edit it; won't care of how the AI reads it.

Now human language, limited as it already is, is to be is to be humbled before machines that humans have also invented. In our inability to create a machine capable of doing cognitively what humans do, we prefer that humans function as if they had been lobotomized, in deference to our crude machines.

We have built god, and god is stupid, and we bow before him because god has been created, once again, in the image of man.

"Our algorithm isn't stupid, it's the literature which is wrong" is not something I expected to read today.

Have you ever read a written passage out loud and failed because the text was too contrived and the nature of the intonation only became apparent after re-reading the sentence multiple times?

I have, plenty of times. And in most of those times I got really angry at the author for writing "wrong".

So yes, to me saying that the literature is wrong is nothing unexpected.

It depends. Is it technical writing? Then I agree with you. But literature is meant to make you feel a certain way, to paint a mood or a picture, or even to experiment with language.

The "AI" has trouble with that because it's fundamentally a machine for recombining already existing information.

It’s more like “the algorithm is stupid, so it can’t fix bad literature in its brain like humans do”

> Unfortunately the AI completely faceplants, placing an enormous pause right in the middle of the phrase "all places wild". It actually changes the meaning of the text, making it sound more like "...the beauty he saw in all places. Wild!"

I disagree for 2 reasons:

1. There's a perfectly fine reason to put a pause between "places" and "wild": to put emphasis on "wild". Bill doesn't see beauty in all places, but specifically in all wild places.

2. Interpreting the narration as "[...] all places. Wild!" is farfetched because the narrator pronounces "wild" very calmly and softly.

I agree the pause is a bit too long, but I was expecting way worse when I read your comment about how "the AI completely faceplants".

I disagree with both reasons.

1. A long pause between "places" and "wild", to me, signals their dis-association, that "wild" does not link with "places". However, the lack of punctuation in the written text implies the phrase "all places wild", the "all wild places" you refer to. I'm with the GP here, the AI didn't convey the meaning I'd expect from the text.

2. Also, the preceding text seems to discuss a certain "ineffability", a spiritual/magic in the world that seems diffuse, broad and subtle. With that context, pronouncing "wild" calmly and softly ties it to the earlier ineffability rather than the more discrete "places". Again, this reinforces the "...places. Wild!" interpretation. I am very impressed the AI used two or more modes of expression (time, tone) to express a feeling... but I disagree that's what the text held.

Maybe the AI's smarter than me.

> A long pause between "places" and "wild", to me, signals their dis-association, that "wild" does not link with "places"

But that long pause is still way shorter than the pauses at actual periods! Look at the waveform[0]: the ovals are the periods, the rectangle is the pause between "places" and "wild". I guess due to the length of the pauses at the actual periods, my brain automatically discards the possibility of "all places. Wild!" and then the best interpretation clearly is "all places wild" for me.

But hey, the fact that at least two people interpreted it differently says something. Maybe this was more of a faceplant than I initially realized.

[0] https://imgur.com/a/j9CtZFZ

Empirical evidence of a pause and its length is not indicative of how the typical listener will interpret it.

Maybe this is just me, but I find it really unnatural to pause in the middle of a short phrase like "all things wild". I'd emphasize "wild" by putting stress on it, not by pausing.

But this excerpt is the end of a paragraph that begins with "Bill loved and found solace in nature." and describes taking walks and looking at the moon. This doesn't support emphasizing "wild" because the author has already established that information; the important part of the sentence is "deep spiritual yearning", or maybe "easiest to express", since the author then goes on to discuss how, after he died, Bill expressed himself from beyond the grave in other ways.

I could kind of understand the AI not quite getting the emphasis right, since that's a judgement call that requires a lot of context from the rest of the book. But breaking up "all places wild" like the sample does suggests that it doesn't understand the basic grammar of the sentence.

I'm guessing that the main reason they require you to go through their "preferred partners" is that their job is to insert annontations that the speech generator needs to make it sound good.

I wonder if this is because it's difficult work, or if the tools aren't user friendly enough to put in the hands of untrained users. If it's the latter I suppose that sooner or later we won't need to go through the partners.

The sample is taken from a full audiobook that is currently available for sale, so you'd think they would've put an annotation on "all places wild" if they had that ability.

Perhaps this bug could be a feature?

  A panda walks into a café. He orders a sandwich, eats it, then draws a gun and fires two shots in the air.

  "Why?" asks the confused waiter, as the panda makes towards the exit. The panda produces a badly punctuated wildlife manual and tosses it over his shoulder.

  "I'm a panda," he says at the door. "Look it up."

  The waiter turns to the relevant entry in the manual and, sure enough, finds an explanation.

  "Panda. Large black-and-white bear-like mammal, native to China. Eats, shoots & leaves."

https://en.wikipedia.org/wiki/Eats,_Shoots_&_Leaves

By normal sentence construction it would be all wild places. It’s a good test as you say, the author is having fun with grammar to give you the idea that it’s a subset of places rather than “wild places”, so I would expect it to be written with a link between places and wild, “all places-wild.”

I've seen this construction used in a lot of places (like the name of "All Things Digital", the predecessor to Re/code), but I have never heard of anyone putting a dash between the last two words.

And might I add, Helena does not sound like a Soprano given the speech tone. (Also, smoking is bad mmmmkay)

A number of books I've been reading aloud leave out the comma after a prepositional phrase (and friends), and it totally throws off my cadence.

Reply on news.ycombinator.com