First of all, real men start counting at zero. Because computers.
Second, math people are not limited to integers. Thus, you’re reading part 1/2 of a story.
Third, dictation software is not as awesome as it seems.*
More about that third thing: I’ve been inhumanly busy lately. In an attempt to streamline blogging, I dictated a very short story while driving. Fail! Our modern world can livestream Facebook onto a teenager’s smartphone but can’t take dictation worth shit! I have a handheld audio recorder that is a wonder. Excellent fidelity, easy to use, great microphone, etc… My truck, despite having the usual Dodge-ish propensity to steer like a pregnant yak, is pretty quiet. Cruising down the road, the microphone recorded every word crystal clear and with minimal background noise. I also have Dragon Dictate, which is among the better dictation options. I even “trained” it to my voice and vocabulary. Yet, when I fed the audio through Dragon Translation I got this:
These masters will recycle saddlebags is to work you a is is a lot of acorns to fill motors saddlebags is a ridiculously large amount of time him acorns you recycle and dropping them in the saddlebags and was, I wish that I had all the acorns of one a couple days later I returned from my trip is in a box set in the kitchen catches his is usually email territory and it is true is curmudgeon insulin
Three pages of that shit! I dictated the story and even I have no idea what I was talking about. It’s like David Lynch did that talking backwards thing to Eliza Doolittle while she spews a Cockney accent and has a mouthful of marbles. There will be a small delay in the story while I take my laptop behind the woodshed and beat it senseless.
A.C.
P.S. Also, I accidentally electrocuted myself while rewiring electrical outlets. I hate it when that happens. While this has nothing to do with acorns and can’t possibly be the fault of either Dragon Dictate or Dodge, I’m blaming them anyway.
* If you’re about to work yourself into a froth about how awesome Dragon happens to be, tell me what I’m missing. I’ve given it a proper unbiased test. If I use a top notch microphone at perfect tune in a quiet office and dictate very slowly and enunciate like I’m talking to a drunk, stupid, toddler, it’s only modestly useful. It’s probably awesome for repetitious tasks like medical records or if you’re injured but it just isn’t as great as I’d like.
You were shocked….not electrocuted…..if you had been, well, there probably would be no further worries with dictation or other activities. And no more curmudgeonly stories….and that would be no good a’tall.
Good point. I inadvertently conflated shocked and electrocuted. Big difference!
I think getting shocked in 110V is God’s way of telling me I shouldn’t be playing with real electricity. I am trying hard to stick with 110V instead of the dark magic of 220V because who knows what a “shock” would feel like on one of those circuits?
I am shocked—shocked—to find that milli-amping is going on in here!
Alternating current is micro-aggressing me!
Used to call it the 60hz bugaloo!
Make you dance at a vibration speed.
Whats the natural frequency of an idiot who doesn’t clear and test the breaker? 60 hurts!
60 Hz Bugaloo was such a great phrase I put it in my next post. Excellent!
So what you’re admitting is that you were stone drunk while you were driving down the highway and using Dragon Dictate? I hope you’re aware that “Law Enforcement” can introduce the “transcript” into evidence at your trial for Driving While Bombed…
Ha ha ha… on the other hand I have ready made evidence for my “Too Stupid To Be Convicted” defense. It’s like the “Insanity Defense” but applies to bloggers.
Just say you were “Tatered”
That would be a great recording. “We found this recording device in his truck. Apparently he had the right to remain silent but didn’t have the ability.”
I gave up on Dragon several years ago. BUT – I needed a solution, and have been super happy for the past 8 months or so using the native voice recorder on my android phone, along with a service called VoiceBase (no relationship, beyond super satisfied customer.) I use this in all sorts of manners and venues – including a super loud car (Toyota FJ with 120k – and rattling like a tin can with pebbles warning tripwire,) in the middle of a conference room, using a headset and capturing both sides of a call, etc. Super cheap for their machine translation which seems to have a better algo than Dragon (might be using the AWS backend?,) and if that quality is not good enough, they offer a slightly less super cheap (still ridiculous value IMHO) human audited transcription. Normal return time after uploading an MP3 of less than 30 minutes in length is single digit minutes (machine.) Also – no super annoying personal training (yeah, I did that with Dragon back when…)
Thanks for the info. I think you’re referring to here–> https://www.voicebase.com/products-features/
I read that it’s $0.02 / minute for translation. Does that mean you send them a 10 minute audio file and for $0.20 you get a text file (that’s not gibberish!). I think I can spring for $1.20 an hour for squirrel stories. I’m not really concerned if it’s a 30 minute return time or takes a couple of days, but I am cheap.
One other thought, there must be different parts of the brain involved in typing rather than speech. I find it massively harder to tell a story to an audio recorder than I do to type it. Probably I change every sentence three times while I type it but when I say the words they’re immediately locked down in the audio. I don’t know if I’ll get “used to” this with time but so far it hasn’t worked.
Yep, exactly. Although (it was months ago) I seem to recall finding a trial offer that gave me an hour or two free to test. I honestly get so much bang for the buck, that I don’t even notice the expenditure. Not to mention that in addition to the time savings, the extra creative productivity gained while being able to dictate in the car/studio/airplane/etc. A huge time saver (and almost more important, the ability to capture creative flow – I do marketing strategy) for me, and dirt cheap to start with.
As to changing and editing – same problem here initially.
What worked for me, was recalling a clerkship I served for some very impressive attorneys 20+ years ago. One of the partners, Richard, was much like us, and I asked (at the time it was ALL human transcription) Melanie his admin how she dealt with it. Richard had told me that once he got in the groove, he just kept talking (it was specifically an exercise he practiced) and would just say “strike that” and just keep speaking his revision. Melanie would listen to the tapes at 2x speed before transcribing, just noting where he was striking on a time log – and would then proceed to transcription anticipating strikes at certain points.
Nowadays, I have found similar to Richard, that talking out an issue to myself while recording is “flow” time, something I am by nature of my gig, pulled out of all to often.
So, if I am lucky (or disciplined enough to) get in flow, then I try very deliberately to stay there uninterrupted.
I do this by imitating Richard, and just talking on and on, often verbally restating an idea 4, 5, 8 times! The tech to transcribe it is so ridiculously cheap (Melanie was incredible, but not cheap,) that I just go in the transcript that comes back, find (ctrl-f) anywhere that says “strike” and going back and cutting out anything prior to that term.
Pro-note, I don’t delete the stuff prior to striking, because I might change my mind about it in the future, I just dump it to a different document as raw notes! That has saved my butt several times!
Thanks. I’ve mentally slated some time after Christmas for writing, I’ll see if your ideas work out for me.
Just how thick is your accent anyway? Adding a couple syllables to each word are we?
Perhaps the Nirvana effect?
“It’s unintelligible, I just can’t get it through my skull
It’s hard to bargle nawdle zouss with all these marbles in my mouth”
Try the Geordie accent (which, in some parts of Britain, is considered a speech impediment):
https://www.youtube.com/watch?v=kayiRqL-nTY
Stick THAT through your speech to text translator! >};o) And, yes, I am a Gerordie.
I’ve found that Android’s speech to text works pretty well, although I’ve learned it works best when I slow down a bit, and also pause a hair between sentences (or even in the middle of long sentences.)
Hm… I don’t have much of a smartphone. I may see if I can run something like that in Linux of a Chromebook. I didn’t have great luck with Windows’ included utility but it was a few years ago I last tried it.
Just be careful with any Alphabet “product” lest you find out that it translated your speech as an order for a $5000 donation to NPR…
That reminds me, I’ve been meaning to link to Louder With Crowder’s experimenting with Alexa.
If somebody says “Choose a number between 1 and 10” I usually say something like “Pi.” Or “4.0026” (the molecular weight of helium.
On the subject of “not being limited to integers.” (There are other, equally obnoxious choices, like the square root of 2)
There is nothing imaginary on my blog; including the parts with squirrels.
I suppose that applies more to i than square root of 2. Fuck it, I’m going to go eat a pi.
I’m with ya’ on that. This also is why math people don’t get invited to parties. (Or at least people learn to stop asking them to pick random numbers.)
AC is math literate! Who woulda thunk it? I was reading your comments to someone when I started to editorialize that the square root of 2 isn’t imaginary but then kept reading.
Yah, I know my math-ology. Don’t tell anyone though. It scares the squares. Best to keep superpowers like calculus under wraps.