Text-to-speech Rendering Sample

A word from our sponsor:

Printer-friendly version
AttachmentSize
Audio icon Easy As Falling Off A Bike pt 29951.11 MB

Author: 

Blog About: 

Angharad asked about the ability of text-to-speech to render British Slang. Lacking any knowledge whatsoever of said British Slang, I offer the attached rendering of the current issue of Easy As Falling Off A Bike pt 2995. The software is an old version of Text Aloud by NextUp.com (very old as software goes), and the current versions are (hopefully) better.

Will repeat for a few issues or texts of your choice.

Comments

Fixed, I hope

erin's picture

I think I fixed the attachment and the link to it. If they still don't work, try force reloading the page to clear your local cache.

Hugs,
Erin

= Give everyone the benefit of the doubt because certainty is a fragile thing that can be shattered by one overlooked fact.

Wow, my little ol' scribbles on audio

Angharad's picture

Thanks for that Sara and Erin, pity it was read by a man's voice and it sounded like he was in a rush, otherwise it was certainly something different. Couldn't distinguish between tenses of read and it messed up vire, but it was interesting.

Angharad

Well, that was... interesting!

I certainly hope other versions of text-to-speech are better than that. I managed to listen to about 15 seconds of it before I had to hit stop. I'm sure I can still feel brain cells ricocheting around my skull.

I did a little test of my own

I did a little test of my own after listening to that less than pleasing rendition of Bike.

I sent Bike 2995 to my Kindle Fire and used the text to speech feature with the British English female voice. The result was rather good, quite satisfactory, really. I don't see any way I can capture it as an audio file to share, so, sorry about that.

A modern text to speech engine with an appropriate voice can do a good job of rendering Bike into the spoken word. You won't be fooled into thinking it's a real person, but you won't want to stick sharp objects in your ears.

Kris

{I leave a trail of Kudos as I browse the site. Be careful where you step!}

Thanks to ...

Sara Selvig's picture

all for the feedback. Old "readers" are definitely an "acquired taste."

When it is the choice between "reading" and "not reading" my favorite authors, I opt for "reading," even if the voice is not as pleasant as the author's!

There are "more modern/more expensive voices and, I think, some that are more tuned to British pronunciation and inflection. Also, NextUp has made at least one major revision since I acquired.

Of the "free/cheap" voices, I find "AT&T Mike" to be the least objectionable, especially when I push the speed up a few notches to maximize my assimilated info per minute. (Needless to say, when I "read" S Hawking or E B Browing, I crank is down a bunch!)

I would not recommend "reading by ear with NextUp" over "reading by eye" unless "reading by eye" is not feasible. Perhaps it is time for me to bite the bullet and get a "modern" reader, even if it means groping through a new GUI!

Sara


Between the wrinkles, the orthopedic shoes, and nine decades of gravity, it is really hard to be alluring. My icon, you ask? It is the last picture I allowed to escape the camera ... back before most BC authors were born.

Years ago,

when Wordperfect for the Mac still existed, the proofreading function was quite good when using one of the female voices. It caused no pain to this recording engineer.

I still miss it as it was much the best way to proofread text.

Liz

Thank you for posting this sample

The following is MY opinion.

It has confirmed to me that I shall NEVER employ a text-to-speech software package, unless I am struck blind - may that never happen!

Frankly, the voice was wrong, the emphases were wrong, the pronunciation was wrong and the artificially inserted pauses, or 'gaps' were annoying. It completely ruined the atmosphere of 'Bike' with no emotions, no sly little double-entendres, no little piquant digs, no 'life' in the speech.

There are so many fascinating characters in 'Bike' but to have their voices all uttered by the same unemotional robotic speech is simply destroying a lot of what is good in the stories.

I frankly cannot understand anyone wanting to listen to this drivel rather than read it themselves and let their own brain fill in the 'Life' of it all.

This is, to my mind, an extension of why so many people are disappointed when a film comes out based on a favourite book. Having read the story, a discerning reader's mind conjures up mental images of the scenes, filling out the experience with fond 'snapshots'. Then the film comes along and it's:

"Oh no. Peter doesn't look like that!"

"That room is wrong, the yellow just doesn't suit."

"She would never be such a whimpering, simpering little mouse."

And so on.

I am going to have to work hard to erase the memory of this travesty, I'm sorry to say.

Thank you for showing me the light.

I shall go and make a cuppa now, and let the shudders subside.

Julia.

Go ahead Julia, tell us what

Go ahead Julia, tell us what you really think. Don't hold back! ;-)

There are more people than you might think, including some here on BCTS, who have varying degrees of vision deficits that make reading a chore, or impossible. For them, some sort of text to speech solution is a must.

As I pointed out, the sample posted does not represent the current state of the art. Now for non-pleasure, non-fiction reading (I.e. work or study) almost anything will do the job of conveying the basic information. But for fiction, for leisure reading, most of us would prefer something more like a human narrator's voice. Short of hiring a professional to do the narration (like Audible does for many books), the latest generation of text to speech engines combined with a quality appropriate voice, are much better than you might expect.

Kris

{I leave a trail of Kudos as I browse the site. Be careful where you step!}

I agree with ...

Sara Selvig's picture

all you say. And, yes, I would prefer to hear it read by Angharad. With that, and staying within the realm of the possible, I prefer a grating, unnatural, impersonal reading rather than being shut out from Bike (et. al.) entirely.

The mindset I adopted when learning to work with this "Michael" was that, relative to the printed word, it does not reduce the emotional or intellectual content. I must provide the emotional content, just as I had to do when I could read "by eye."

By comparison, the criticism by most appears to be that "Michael" does not add the content that an accomplished human narrator would add.

All that said, I am sure that there is software that would do the job more gracefully. Perhaps I can find demos of them on YouTube. And perhaps I can find one with a usable interface so I can join y'all in the 21st century!

Sara


Between the wrinkles, the orthopedic shoes, and nine decades of gravity, it is really hard to be alluring. My icon, you ask? It is the last picture I allowed to escape the camera ... back before most BC authors were born.

Most Windows apps with text

Most Windows apps with text-to-speech ability use the speech API that comes with it.
Some install additional voices, other just use what's there. Windows ships with one voice, male in XP, female since.
Interestingly, even the apps that ship without, can usually use voices installed by other apps.