Re: Self-Produced AudioBook with NVDA


JM Casey
 

Interesting discussion here and I appreciate you explaining howy ou did this.

 

I’m sort of jumping in late to this thread – but yes, human narratino is 100% the way to go as far as I’m concerned. I would not listen to a TTS read poetry, personally, unless I absolutely had no choice and was required to study it for school or something. I don’t even like listening to a TTS read good prose. It doesn’t matter how “human-sounding” it is or claims to be, it’s just not what I want from the literary experience. This is why I do all my reading on a braille display n owadays, when possible.

If one can’t narrate themselves, I would say, pay someone to do it, or maybe you have a friend with an awesome voicea nd a good understanding of your work who will help for nothing more than a couple of pints.

 

 

 

From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Giles Turnbull
Sent: November 7, 2021 11:02 AM
To: nvda@nvda.groups.io
Subject: Re: [nvda] Self-Produced AudioBook with NVDA

 

In theory it is ... sort of! The main concern is the sound of the TTS engine which, as others have mentioned, might not be very pleasant to somebody not used to listening to screen readers. when my poetry pamphlet won a competition and was published in 2017, it was important to me to have an audiobook version in case any visually-impaired people attending my readings and launch events wanted to buy a copy.; I did a reading at Cardiff Institute for the Blind (now Sight Life) in addition to literature festivals in Abergavenny and Torquay, and several launch events around Wales.

I can now memorise between 30 and 45 minutes of my own poems, but at the time of publication I could only manage about 10 minutes. I tried using Orcam glasses to read from the pamphlet or printout, but the result was too variable to risk using it in front of an audience. I decided to use NVDA on a tablet with three small external Cube speakers which provided plenty of volume to fill an area large enough to hold 30-or-so people ... my first event was at Waterloo Tea, a cafe in the wyndham Arcade in Cardiff. I think I did half a dozen poems from memory and used NVDA to do six or seven more.

As others have mentioned, the screen reading software needs a bit of a helping hand. I edited text files of the poems so that NVDA paused where a human reader would pause, even if it made no grammatical sense to have a comma at that place. I use the Microsoft SAPI5 voice Hazel for public audiences, because it sounds very human compared to any other TTS voice I've heard .. I have a little piece that Hazel reads before her last poem, and she always gets applause from the audience :)

On the morning of my first launch event, I was pondering how to get Hazel to say the phrase "power suits" in a less plummy voice though. The problem was that Hazel pronounces suits see-oots. I tried several variations ... the soo sound could be spelled Sue or Soo, but there is no ts sound in English and joining ts to soo produces soots, and attaching it to Sue produces suets (suet ebing the fat surrounding the kidney!) ... I even tried Sioux but again still the same problem because joining ts to Sioux sounds like the original see-oots. Two hours before I set off for the event, I wondered whether changing the consonant from a t might work ... and indeed, writiing suits as Siouxds worked (with the Hazel voice) almost indistinguishable to suits :)

When I produced my audiobook version I paid £100 for half a day of studio time with a an engineer. He micéd me for the poems I did from memory (by this time I culd do 10 poems in the pamphlet from memory) and then he micéd my tablet for Hazel to do the remaining 10. Was it worth doing the audiobook? Financially speaking, no! I have sold two copies of it so far (I sold them at events on USB thumb drives for the same £5 fee for the audiobook ... bearing in mind I buy my copies at 50 percent) so I'm currently about £95 down on my investment in the audiobook ... but it's worth it for me to have done that for the visually-impaired readers :)

Join nvda@nvda.groups.io to automatically receive all group messages.