Dictator... walking along a hiking path dictating plot notes or story content into a digital recorder. Back home, uploading the voice file for transcribing, either by an online service or into a speech-to-text translator.
Some authors claim this increases output, overcomes physical problems such as carpal tunnel affliction, and as a side benefit, gets a chair-bound writer out into fresh air and exercise.
Comments? Any experience with on-board software such as Dragon NS?
I've used the Dragon Professional version with MacBook Pro for writing reports. It's good, but you'll still end up editing.
I've used the Dragon Professional version with MacBook Pro for writing reports. It's good, but you'll still end up editing.
That will likely happen with any transcribing software, as they're notoriously unreliable--especially where numbers are concerned. Also, graybeard will need to learn to speak punctuation to ensure his recordings read correctly, the same way most Siri and Cortana users have to when transcribing emails on their phones and/or watches.
I've just discovered a non-time-limited recorder for my new Apple Watch, so I'll have to try it out for reliability soon.
I was involved with this stuff years ago in the corporate world, when we were designing trading floors on Wall Street, and trying to convert the online stock feeds into publishable online documents, and the quality of transcriptions haven't improved that much, even given how much time has passed. Instead, software designers focus on getting apps to recognize Zwalheli and heavy Scottish accepts rather than ensuring what's spoken is accurately captured. :(
I've tried that before. Dragon worked fine for getting the words in, but there was so much editing, that for me, it's easier to just type. We've had discussions about this before on here, as some of our authors have physical handicaps that they can't just type things out, and it works for them.
Dragon worked fine for getting the words in, but there was so much editing, that for me, it's easier to just type.
Yes, that's a concern. It seems the 'fix' for that is to train the software to eliminate the mistakes. Users also report that a high-quality microphone is required for clean results. Reports I've seen claim a 1 in 100 error/word ratio, which I'll cheerfully accept. The intent is for garbage-drafting. Whether speech-to-text or keyboarded, that first draft will need a LOT of editing. Auto spell check should also highlight many of the errors.
I've also learned the Mac version is crap, and has finally been discontinued. The Windows version is highly trainable. Only 'premium' or 'professional' versions will accept audio file input from a digital recorder. Here's hoping... a copy of DNS Premium will land on the homestead porch next week.
Never had a problem with the Mac version. I'd be curious to know where you found that information?
Never had a problem with the Mac version. I'd be curious to know where you found that information?
The Mac was long the Gold standard to text-to-speech apps, with most of it handled on the system level, which Windows, as usual, relied on 3rd party vendors producing a variety of incompatible software which wouldn't work with anything else.
Unfortunately, Apple has never really upgraded their now decades old speech engines, and they're showing their age. They still integrate speech into their OS much better than Windows does (except for system startup, when much of what happens is hidden under lays of hardware encryption to prevent system hacks).
It would be nice to see those upgraded, but I'm quite holding my breath waiting for it a long time ago!
Unfortunately, Apple has never really upgraded their now decades old speech engines, and they're showing their age. They still integrate speech into their OS much better than Windows does (except for system startup, when much of what happens is hidden under lays of hardware encryption to prevent system hacks).
Do you have a source for that information I can research?
So you know, when I ask a question like that, it's not in doubt, it's because I simply don't know. I've retired in the last year and I'm finding some holes in my information base that irritate the hell out of me. This is the sort of thing I should have known, but somehow missed along the way.
Do you have a source for that information I can research?
Alas, all of my research is from decades ago, which are now long lost, though I haven't noticed (on my own observations) that much, if anything, has changed. The few that I've tried aren't any better than those decades old, and few of the 'old guard', like Dragon Naturally Speaking, haven't had any substantial increases in a long time either.
Again, they all adapt to the latest changes in OS additions, but no one wants to rewrite their base coding, as they already have all the customers they expect to get. :( Essentially, it's a stagnant industry. The little research being done is in listening (mostly to casual comments about products those listening can then market to them), rather than making the underlying products any better.
On a side note, the products which I had almost 30 years ago (with a now defunct computer company) were vastly suprerior to what's available now. Back then, we had tools to rapidly craft tools to parse and decode language, whereas now, if it's done at all, you've got to code everything yourself from the ground up.
In my day, I could slap together a new language translator in a matter of days, now ...
Without the backing of an industrial base, and with no apparent sign of huge, hungry market, no one's investing anything at all into the industry (which I guess also includes me, since I actually know better).
Still, what tools are still available are all roughly equivalent. You'll need to speak fairly slowly and clearly, you need to say punctuation in order to enter it (ex: "This is a full stop, period."), and you'll need to waste quite a bit of time editing the garbage that results. It still works, it's just not any better than what we struggled with years ago, and does add to your editing time (think 'pre-edit' edits).
Still, what tools are still available are all roughly equivalent. You'll need to speak fairly slowly and clearly, you need to say punctuation in order to enter it (ex: "This is a full stop, period."), and you'll need to waste quite a bit of time editing the garbage that results. It still works, it's just not any better than what we struggled with years ago, and does add to your editing time (think 'pre-edit' edits).
The information I've picked up is from successful published authors who use the dictate method to draft their novels. (Key point: "draft," needing editing & revision) They seem to prefer a digital voice recorder while on the move (one while commuting an hour each way.) Another prefers to get outside, and dictates while hiking. Others suffer carpal tunnel problems and can't type lengthy drafts.
The key point all of them seem to make is that DNS has improved its error correction and training process to the point that patient and dogged error correction and training for their distinctive accents or voice patterns does result in what they claim is a very low error rate, as high as 99 percent accurate. Some of these authors are UK-based, and I'm here to tell you, listening to them on YouTube is an exercise in futility. I have to click the "CC" closed caption button to see their spoken words; my ear sure as 'ell can't translate their dialect! So if DNS works for them, I'm confident it will work with my elder-raspy voice.
The biggest attraction for me is the ability to wander my favorite state park, dictating into a voice recorder. I did that years ago recording observations while truck driving through America, and I found it extremely frustrating. An hour of voice recording would take three hours or more of transcribing on the keyboard. Not workable, at all.
So DNS and current OS tech might not be as good as it could be, but my research has convinced me it's worth trying. Another point: all sources I've read say it's a hard task to get used to "writing" by story-telling into a recorder. And that ain't no big surprise. I'm old enough to remember it took me a few weeks to get comfortable story-telling (news articles) while pounding a typewriter keyboard. It was NOT an easy transition from pen & paper.
Anybody old enough to remember that?
The key point all of them seem to make is that DNS has improved its error correction and training process to the point that patient and dogged error correction and training for their distinctive accents or voice patterns does result in what they claim is a very low error rate, as high as 99 percent accurate. Some of these authors are UK-based, and I'm here to tell you, listening to them on YouTube is an exercise in futility.
That was my earlier point. Most of the more recent upgrades have been in 'adaptive integration', or software that learns to recognize your speech patterns. The underlying transcription is basically pretty strong (they've always been good at conveying plain text, where they're weak is at accurately recording complex numbers—but that's more about OCR than dictation apps).
I'm old enough to remember it took me a few weeks to get comfortable story-telling (news articles) while pounding a typewriter keyboard. It was NOT an easy transition from pen & paper.
Anybody old enough to remember that?
I've always told people that 3 best courses I've even taken were: 1) typing 2) speedreading and 3) Calculus.
Since I took typing way back in early high school, I've long typed at 300+ words/minute (which is why my messages here are always so long!), so for me, that's never been a minute, as I can type much faster than I can speak aloud—especially if I also have to speak basic punctuation marks.
The speed reading I've actually let go, as it was negatively impacting my ability to recognize typos (speed reading is based in guessing what a sentence says by the shape of the words, so you tend to assume which words are there, rather than wasting time reading each one). While that process has killed my pleasure reading, it's made me a somewhat better self-editor.
Calculus, though, that's a whole 'nother story.
It still works, it's just not any better than what we struggled with years ago, and does add to your editing time (think 'pre-edit' edits).
Thanks for the reply, I hadn't given any thought to the amount of editing I was doing until this morning.
Thanks for the reply, I hadn't given any thought to the amount of editing I was doing until this morning.
It's basically a trade off. An extra step (pre-edit edits) on top of your already onerous editing requirements. For some, especially those with visual issues slow typing speeds, it's a sensible move, but for most, the extra step takes considerably more time.
Unlike me, my bother still believes that smartphones are simply 'idiot boxes on steroids' and has never learned how to type. Instead, he writes everything out three separate times, before turning it over to his poor wife to transcribe, which is a miserable task.
I'd tried to get him to use Siri Dictation (not the most effective app, but much better than what he was doing), but he quit after only two sentences. He takes pride in his Luddite tag, but at least he's now comfortable with an iPhone--even if he has no desire to learn many apps.
Never had a problem with the Mac version. I'd be curious to know where you found that information?
I came across several of those comments from users who recommended very strongly to use the Windows version under Parallels or some other emulator on the Mac, rather than the Mac version. The fact that the Mac "Dictate" version has been discontinued sorta/kinda speaks to the issue, I'd think.
I came across several of those comments from users who recommended very strongly to use the Windows version under Parallels or some other emulator on the Mac, rather than the Mac version. The fact that the Mac "Dictate" version has been discontinued sorta/kinda speaks to the issue, I'd think.
I was working on the assumption, if it ain't broke, don't fix it. It did what I needed it to do. However, after the comments here, I began reading to it out of a Heinlein novel. I hadn't realized how much editing it was requiring until this morning while doing that.
We've had discussions about this before on here, as some of our authors have physical handicaps that they can't just type things out, and it works for them.
As they say, you can't edit a blank page, so whatever gets the words on the page allows authors to fine tune the words.
Amazing! I first tried Dragon MSDOs Voice to text in 1985. I was amazed [being a techie] at how well it worked but still realized it was much slower than just typing it.
Judging from how 'Text Suggest' and autocorrect works on my phone, it probably still is, lacking a full blown AI computer.
I'll bet it will work near flawlessly in about ten years. Online cloud APP and high bandwidth, of course, for a price.
Sidenote: I flunked out of typing in HS [teacher asked me to quit after 2 weeks] . Shortly after that joined the Navy and got a billet where I HAD to type. Learned hunt/peck pretty quick. But never enjoyed it till later I used an IBM selectric. Like magic.
Dictation is not only slower than typing, I speak in garbage grammar. I even tried dictation and having an exgirlfriend type it out. She wasn't an ex at the time. It was a bad move. On a keyboard I can do acrobatics, monologuing, I can hardly walk straight.
My new copy of DNS Ver. 13 arrived today, with included mic headset, adapters, and a digital recorder that outputs .mp3 files.
Installation was weird: the DVD thrashed at least two years' life out of my drive, but after half an hour all was installed. A brief setup and training session followed. Color me overjoyed! The live voice transcription is very, very good thru a mic plugged into the computer. It's like magic to these old eyes.
The acid test followed. After another short training session under a different profile on DNS for recorder transcription, and once I'd learned which of the 36-odd superfluous 'features' I could ignore on the little digital recorder (no, I don't need the radio feature!) I read a few pages of a story into it, and fed the file to DNS. Excellent! Superb!
Perfect? Not hardly, but -- boy, howdy! -- being able to ride or hike around and talk notes and scenes and story stuff "on the fly" into a shirt pocket recorder, and come home and feed it into the computer for transcribed printouts: fantastic!
I find the accuracy is more than okay. After all, everything keyboarded has to be proofread and edited anyway, and I find the DNS 99% good is quite acceptable.
All the normal weirdness is present: to, too, two are a crapshoot; so is other arcane content and punctuation. Some can be worked around; other glitchy stuff, we live with and correct with edits.
For a $77 buy of an earlier version (15 is current) and a nice recorder, I'm happy. The included headset is 'shite' as Brits would say; it's barely adequate. But I'll hardly ever use it, prefering the recorder for 'walk-about' sessions.
So: I think I'd best avoid using it on the WA state ferry rides, lest the Homeland Security undercover folks think I'm casing it for... well, anyway, lest this be scanned for the So. Utah memory banks, we'll just avoid saying and not talking into my hand on the ferry for nervous ninnies to report. Kidding? Not hardly. Take pictures 'looking out' okay; take pictures 'looking in' not so okay. Get interviewed upon arrival. So... maybe pretend it's a cell phone? Yeh... that's it! Talking to the cellphone in my hand, Okay.
eBay, actually. Search for DNS Ver.13 Premium, and look for the pix of the carton with the recorder. It was $77 "Buy It Now" from a California seller.
FYI, the "Home" version won't transcribe external files; and the "Professional" version is overkill for private users.
(I gave Amazon first chance, but four days after ordering it still hadn't shipped, so I cancelled. eBay seller had it to me in four days. Same price. Free shipping.)
It was $77 "Buy It Now" from a California seller.
I looked at availability through the local sellers here in Australia and the cost varied from A$250 for DNS Home to A$500 for DNS - no home or pro versions in the name. For the pro versions they can go up to A$2,500 and no microphones with them either. based on the recent conversion rates US$77 should come out at about A$150 or lower. Danged if I know where the other hundred bucks is going.
Sorry to hear that, EB. FYI, the DNS retail box states the product is "Not For Sale in Europe." Curious.
As a side note, it requires 'validation' through a 'net connection, so a US version sent to OZ might not validate? Also, the printed materials stated it was licensed only for one machine, within one physical address. But I was able to validate it on the Desktop box, and the laptop.
That listing has the headset but no digital recorder. The packages with the recorder have an illustration of the Philips recorder on the front of the DNS box. BTW, the recorder which came with my package is the Philips DVT4100, which lists separately on eBay for $40.
Dictator... walking along a hiking path dictating plot notes or story content into a digital recorder.
Of course 'Dictator' is his nom de plume. His real name is Richard Potato.
Yeh, he's got his eyes on you. Best to keep yer eyes peeled lest trouble sprout.
Mash good it'll do him. If he peeps whilst I'm sautéing he'll get a kick in his rösti, jackets off and when the chips are down... His kids will call him parboil.
Your move... :)
I think we've strained and rehashed this; now 'tis overdone to a crispy fritter.
Speaking of spuds, there's an absolutely delightful solo Canuck from Alberta who's something of a roving romantic on YouTube: Slim Potatohead. Wanders about in a Jeep SUV pulling a folding A-Liner camp trailer. Dunno how spuds got to be his moniker, but I gorantee (like my Cajun idjit cousin sez) that you've never seen prettier homemade videography of the boondocking places still left in the US, all far off the beaten track, with commentary that iffen it was writ in words, would bring tears to drip on yer page.
Check it out. Slim Potatohead. YouTube.
(No vegetables were abused in the making of this post. Well, hardly any. Not much. Do parsnips count?)
(No vegetables were abused in the making of this post. Well, hardly any. Not much. Do parsnips count?)
You evil parsnip molester! Have you no sense of compassion?!
You evil parsnip molester! Have you no sense of compassion?!
Calm down, the potatoes just sat around taking the piss out of it. You know, a parsnip roast...
:)
