Stay up to date with notifications from The Independent

Notifications can be managed in browser preferences.

Now you're talking ...

IBM's voice recognition software put to the test

Sunday 14 May 2000 19:00 EDT
Comments

Your support helps us to tell the story

From reproductive rights to climate change to Big Tech, The Independent is on the ground when the story is developing. Whether it's investigating the financials of Elon Musk's pro-Trump PAC or producing our latest documentary, 'The A Word', which shines a light on the American women fighting for reproductive rights, we know how important it is to parse out the facts from the messaging.

At such a critical moment in US history, we need reporters on the ground. Your donation allows us to keep sending journalists to speak to both sides of the story.

The Independent is trusted by Americans across the entire political spectrum. And unlike many other quality news outlets, we choose not to lock Americans out of our reporting and analysis with paywalls. We believe quality journalism should be available to everyone, paid for by those who can afford it.

Your support makes all the difference.

Voice recognition has she [for] years been a holy Grail for the computer industry. I dictated that last sentence and a computer tied to [typed] it, which is why the grammar is peculiar. But what's the point in testing a piece of voice-recognition software if you don't get it to do the heavy lifting?

So in what follows, the machine's version appears in plain text - or italics when it gets it wrong, with my own corrections inside square brackets. It correctly spelt my name, by the way.

The software in question is IBM's ViaVoice product, which costs £69.99 in the stalls [in the stores] and which the company insists has been doing gang Buster's [gangbuster's] business. And I can see the attractions. It comes with its own headset and microphone. It has a 100,000-word vocabulary and can learn 60,000 specialist ones. It gets better at learning your pronunciation. You can set it up for multiple users. For people unfamiliar with typing, being able to talk and see it exterior [appear] is far more attractive than picking your way about Turkey where teach [a Qwerty] keyboard.

Doris [ViaVoice] can do that: for people used, as IBM puts it, to "talking up to the full stop" - Batty's, [that is,] composing in the [their] head the entire sentences to be written before they speak - it will do the job very well. For most of us, though, who ought [might] I think a [be] more prone to write something down and then revise it, I suspect the benefits might be less clear at first.

Initial impressions were good. Though the manual is dire and not particularly enlightening, the program installs itself efficiently, and the built- in help and "set-up assistance" do a better job of explaining the software to you than the manual does.

The systems are just [suggests] that a ticket [you take it] through a "voice training " system in order for it to get used to your intonation: this consists of reading any or all of course [four] stories, each taking between 15 and 60 minutes. It's not a burden, and you get an idea of how quickly the machine can catch up with your speech. For a sentence like the previous one, it took about four seconds for the machine to finish processing the sentence after I stopped speaking. The short of [shorter the] sentence, the short of [shorter] the delay. I was using a year-old appal [Apple] power but [Powerbook] cheek three [G3], the lowest specification that IBM recommends. There were some problems: I had then almost completed this review winner of [when] the program inexplicably crashed - before I had saved the text. This happened two or three more times.

I would guess that the speed of recognition is about as good as a competent typist or probably better. And it certainly doesn't misspell com [common] words as I do. You can also teach it special words or phrases that you regularly use. You can thus give it a specialist vocabulary and a Quint [acquaint] it with your peculiarities of pronunciation.

IBM ViaVoice works by analysing words in groups of three, looking for context: once you've worked out the word before and after a middle one, you have a clearer idea what thatone actually years [is].

The trick is to do this with continuous speech, which is why ViaVoice has only recently come on to the market. Programs to analyse discrete speech - where... You... Talk... Like... This - have been around for some years, but they are not satisfactory for everyday use. IBM ViaVoice suddenly [certainly] fits the bill: I think that as anyone who is disabled, or art [that] dislikes typing, would find it a Buna [boon]. But don't think that you'll be able to get away without any typing. Corrections, as you can see, are not a thing of the past.

ViaVoice Standard £39.99

ViaVoice Web £59.99

ViaVoice for Mac £69.99

ViaVoice Pro £119.00

System requirements:

PC: Windows 95, 98 or NT 4.0, 233MHz with MMX, 48MB RAM (64MB for Windows NT), 340MB disk space, CD-ROM drive, sound card. Mac: 233MHz processor, OS 8.5.1 or better, 64Mb RAM

Website: www.ibm.com/viavoice/uk

Join our commenting forum

Join thought-provoking conversations, follow other Independent readers and see their replies

Comments

Thank you for registering

Please refresh the page or navigate to another page on the site to be automatically logged inPlease refresh your browser to be logged in