Using Ancient Voice Recognition Software from 1996! (IBM VoiceType)

[@MichelHoogervorst]

Loved the training montage! 😂

[@DOSStorm]

haha thanks!

[@pianoman4Jesus]

My company was an IBM partner back in the 1990's and into the early 2000's. VoiceType came built in to OS/2 Warp 4. Then for Windows, IBM further enhanced the VoiceType product line. I called on some legal and medical client sites and demonstrated it.... none made purchases.

[@DOSStorm]

Thanks for the inside knowledge! It makes sense. If it worked in a more natural manner it would have probably flew off the shelves, but with all the work you need to get it going and the high cost it's hardly worth it.

[@Vanessinha91Pucca]

I remember in 1999 i went to a school colleague to make a group school essay. He had this voice to type software he used on Word on his main PC (prob this software).. and well, i remember that i on his secondary PC was typing and finishing much faster as he had to repeat many words to the program to get it right

[@remghoost]

Hey, neat channel. Just got you in my recommendations after binging various Computer Chronicles episodes. That "training" montage definitely earned a subscription. haha. Something I noticed around the 11:10 mark is that the software was picking up the mic as "clipping" (in the red). Tweaking the input volume to put it in the green might yield better results....? But, you used it exactly as someone who purchased the software would've used it. I definitely blame the included microphone and janky Windows 95/98 soundcard drivers, not you. And I'm really curious if the software would have an easier time with a modern microphone. I remember those beige stand mics being absolutely horrid for audio quality. Probably wouldn't warrant another video, but I'm still curious if a modern "good" microphone would allow the dictation software to accurately get your words a bit better. Anyways, good stuff! Keep on being awesome.

[@DOSStorm]

Thanks for the sub, I feel like I'm in good company with the likes of the old Computer Chronicles episodes! As for the audio levels the software has a utility to automatically set everything to optimal settings. I did a few recordings in the Windows sound recorder as well and it was actually fairly clear for one of these types of mics. The levels seemed a bit on the low side if anything, and I'm not sure the readings on the software itself are all that accurate. I'm going to try to upload the CD image to Archive.org this weekend if you're interested in trying it. I wonder if it works on Windows 10/11?

[@SeñorDossierOficial]

8:39 so you are a Dink Smallwood fan!

[@DOSStorm]

One of my favorite shareware titles!

[@SeñorDossierOficial]

@@DOSStorm mine too :3

[@Raphipod]

Nice and informative video! What's the name of the song playing directly after the Intro Skit @0:35?

[@DOSStorm]

It called "First Flight". Its from a royality-free synthpop music pack by Ovani Sound.

[@JayTheComputerGuy]

Fun Fact: The Voice Of Demoman From TF2 (Gary Schwartz) voiced Colonel Ilgwom in the JungleBook Game?

[@DOSStorm]

That is cool, thanks for sharing.

[@Vanessaira-Retro]

LMAO! Yeah I remember these programs. Thought these programs were so neat, but then I learned about all the training that was needed to just get it going and even then it was always messing up. Also that MWAVE! XD

[@DOSStorm]

I never used them back in the day. Its a shame they don't work very well, but its an interesting footnote in the history of something so common in everything now. On another note, I hope to cover the MWAVE in more gruesome detail sometime soon. XD

[@DeadCat-42]

I hit the TAB key, but I'm still thirsty.

[@DaarkCloud]

I'm spoiled by M.Modal at work these days. I rarely use it, but it's amazing when I do.

[@DOSStorm]

Yeah, speech recognition is pretty legit these days.

[@DaarkCloud]

@@DOSStorm It's crazy to think that Microsoft now owns this branch of IBM. But what don't they own at this point xD

[@DOSStorm]

@@DaarkCloud I guess I didn't know that but it makes sense. Microsoft seems to really be pushing hard with implementation of AI with speech and video.

[@pc-sound-legacy]

Very entertaining😅 I had a similar software from IBM (ViaVoice?) and my experience wasn't good either. It worked when speaking very slow and pronouncing very clear - but you can type the text faster on the keyboard so there was no benefit at all. Also I don't think it ever worked well in a noisy workplace environment. But these were the 90s and everything with the batch "multimedia" on it was cool, right? At least it was cool enough to impress your mates. I paid way less for my copy (maybe it was bundled with other hardware - can't remember but it sure wasn't several hundreds)

[@DOSStorm]

haha Thanks! I think ViaVoice was a later version and it was more freely available than this software was. Its a shame it doesn't work very well but it is satisfying when it does.

[@autingo6583]

although i'm autistic, it took me to the puddle to understand the training montage. too accustomed to machine learning terms! :-D btw. voice type 3 was also my first exposure to dictation software back in the days. oh, and the error is mostly with you in the "fast speaking" demo, you have to keep at least a miniscule pause. the software for instance always errors out when you forget to set the so-called "glottal stop", an important phonetic component that is not reflected in script (yes i'm also a linguist). here in germany and with the german version of voice type this limitation wasn't so severe because the (glottal) stop between words is a prominent feature of the german language, it is the feature that makes spoken german sound like a machine gun fire to foreign ears, as they tend to say.

[@DOSStorm]

I was hoping everyone would like the silly training montage, I probably wasted too much time shooting it. lol Interesting how it might have worked better for different languages. It makes sense considering the English language has many odd quirks that would pose a challenge for an archaic system like this.

[@autingo6583]

@@DOSStorm it's just that it's a so-called discrete speech recognizer and you have to treat it like this, always keeping micro-pauses between words. takes some getting used to. funny thing is, with the following continuous speech recognizer viavoice, quite a big leap, those micro-pauses you just trained yourself to respect actually worsened the outcome! :-D oh and loved the montage, it was definitely worth the effort!

[@dastrove]

This gives me AVGN vibes

[@DOSStorm]

lol maybe a little.

[@canlelola]

Do not forget, IBM is gay, they even have a song for it! Still it is nice to see how things came to be.