# voice recognition software

**URL:** <https://boards.straightdope.com/t/voice-recognition-software/207081>\
**Category:** Factual Questions\
**Created:** [October 13, 2003, 1:09am UTC](https://boards.straightdope.com/t/voice-recognition-software/207081 "2003-10-13T01:09:16Z")\
**Posts on this page:** 10\
**Page:** 1

<div class="post-metadata">

**Author:** ![handsomeharry](https://avatars.discourse-cdn.com/v4/letter/h/e9c0ed/32.png) [@handsomeharry](https://boards.straightdope.com/u/handsomeharry)\
**Post date:** [October 13, 2003, 1:09am UTC](https://boards.straightdope.com/t/voice-recognition-software/207081/1 "2003-10-13T01:09:16Z")

</div>

i want to speak into a microphone and have my computer type it into word for me. is VRS the thing that i need? does voice recognition software really work? it really sounds like some kind of crock to me. also, it’s so cheap (ibm voice 9 or something like that is less than 30 bucks) i can’t see it being able to take the place of some kind of stenographer.

---

<div class="post-metadata">

**Author:** ![billy](https://avatars.discourse-cdn.com/v4/letter/b/ecd19e/32.png) [@billy](https://boards.straightdope.com/u/billy)\
**Post date:** [October 13, 2003, 1:36am UTC](https://boards.straightdope.com/t/voice-recognition-software/207081/2 "2003-10-13T01:36:26Z")

</div>

I bought some about 7 years ago and it worked ok. I imagine it works much better now with faster processors and 7 more years of algorithym work.

The only real draw back of the stuff I had was that it had to be really quiet for it to work. That means it wouldn’t work well in a cube farm but should be fine in your own home. I wanted it so I could work on reports in the car but it was worthless there. I’m sure that’s less of an issue now.

---

<div class="post-metadata">

**Author:** ![Bill\_H](https://avatars.discourse-cdn.com/v4/letter/b/a5b964/32.png) [@Bill\_H](https://boards.straightdope.com/u/Bill_H)\
**Post date:** [October 13, 2003, 2:22am UTC](https://boards.straightdope.com/t/voice-recognition-software/207081/3 "2003-10-13T02:22:59Z")

</div>

I tried it a while back. It claimed being 95% accurate, and that was likely true. The thing is, when you use it in practice, 5% of everything being wrong is a lot. Like an error in every couple sentences. And unlike typing errors, often these errors were based on sounds, two words combined into some third totally unrelated word or the like.

Someday it’ll be big, but I don’t think it’s cooked yet.

Also, something I really hadn’t counted on: When I use the computer typing, I often don’t think in a linear fashion. I’ll often go back and change words in previous paragraphs, rearrange sentences or paragraphs. And of course that’s extremely difficult with speech.

---

<div class="post-metadata">

**Author:** ![Bob55](https://avatars.discourse-cdn.com/v4/letter/b/e9c0ed/32.png) [@Bob55](https://boards.straightdope.com/u/Bob55)\
**Post date:** [October 13, 2003, 2:32am UTC](https://boards.straightdope.com/t/voice-recognition-software/207081/4 "2003-10-13T02:32:15Z")

</div>

I too tried this about 3 years ago. I bought some software and a headset off Ebay for ~$18. I evidently had to say about 5000 words for it to configure properly - I gave up at about 200! Like **billy** I found that I had to be too quiet, and it also felt weird talking to myself.

---

<div class="post-metadata">

**Author:** ![Bob55](https://avatars.discourse-cdn.com/v4/letter/b/e9c0ed/32.png) [@Bob55](https://boards.straightdope.com/u/Bob55)\
**Post date:** [October 13, 2003, 2:34am UTC](https://boards.straightdope.com/t/voice-recognition-software/207081/5 "2003-10-13T02:34:59Z")

</div>

Oh and I found that I had to talk way too slowly and very concisely - it was just easier, quicker, and more accurate to type.

---

<div class="post-metadata">

**Author:** ![handsomeharry](https://avatars.discourse-cdn.com/v4/letter/h/e9c0ed/32.png) [@handsomeharry](https://boards.straightdope.com/u/handsomeharry)\
**Post date:** [October 13, 2003, 5:26pm UTC](https://boards.straightdope.com/t/voice-recognition-software/207081/6 "2003-10-13T17:26:25Z")

</div>

thanks for the input!

---

<div class="post-metadata">

**Author:** ![aaslatten](https://avatars.discourse-cdn.com/v4/letter/a/f14d63/32.png) [@aaslatten](https://boards.straightdope.com/u/aaslatten)\
**Post date:** [October 13, 2003, 7:15pm UTC](https://boards.straightdope.com/t/voice-recognition-software/207081/7 "2003-10-13T19:15:57Z")

</div>

Like others said: it’s neat how far the technology’s come, but it’s still in the “novelty” phase. It’s still far more efficient to type unless you have some kind of disability.

---

<div class="post-metadata">

**Author:** ![umop\_ap\_sdn](https://avatars.discourse-cdn.com/v4/letter/u/dbc845/32.png) [@umop\_ap\_sdn](https://boards.straightdope.com/u/umop_ap_sdn)\
**Post date:** [October 13, 2003, 8:58pm UTC](https://boards.straightdope.com/t/voice-recognition-software/207081/8 "2003-10-13T20:58:17Z")

</div>

Funny story about speech recognition: I was on a Star Trek message board about 5 years ago, and an actor from the shows was said to be using a headset & voice recognition. When he said Roddenberry (sp?) it got transliterated as Rottenbury.

---

<div class="post-metadata">

**Author:** ![P.Nym](https://avatars.discourse-cdn.com/v4/letter/p/3d9bf3/32.png) [@P.Nym](https://boards.straightdope.com/u/P.Nym)\
**Post date:** [October 13, 2003, 9:15pm UTC](https://boards.straightdope.com/t/voice-recognition-software/207081/9 "2003-10-13T21:15:31Z")

</div>

I had good results with this about 2 years ago. Using the right microphone makes a big difference.

The computer made a few typos, but I was only using it to transcribe notes. The quality didn’t need to be perfect.

Unfortunately, I couldn’t get this to work well with tape-recorded speech. I never figured out how to get the noise-cancelling microphone hooked into a tapedeck. I know this doesn’t sound very exciting, but at the time I was spending over 3 hours a day in the car. Being able to dictate during all that driving would have made me insanely productive.

---

<div class="post-metadata">

**Author:** ![xash](https://avatars.discourse-cdn.com/v4/letter/x/c6cbf5/32.png) [@xash](https://boards.straightdope.com/u/xash)\
**Post date:** [October 14, 2003, 5:09pm UTC](https://boards.straightdope.com/t/voice-recognition-software/207081/10 "2003-10-14T17:09:07Z")

</div>

I’ve tried IBM’s ViaVoice ( [http://www-3.ibm.com/software/voice/viavoice/](http://www-3.ibm.com/software/voice/viavoice/) ) and Scansoft’s Dragon Naturally Speaking ( [http://www.scansoft.com/naturallyspeaking/](http://www.scansoft.com/naturallyspeaking/) ) previously. Although they both do a good job, it’s not anywhere close to the accuracy you would like to have. My experience is from a few years back, and certainly things should have improved since.

I now use Flashpeak’s IntelliComplete ( [http://www.flashpeak.com/icomp/](http://www.flashpeak.com/icomp/) ) While not a Voice-to-Text software, it enables me to enter text at insanely fast speeds, and takes just a few minutes to configure. It has drastically cut down my document and coding typing times.
