Slashdot Mirror

Costs for hardware substitution is not found only on the linux side.

How about PaperPort 6.5?

Note: PaperPort 6.5 is not supported on Windows XP.

How about Epson GT-6000?

This is a parallel scanner, parallel scanners are not supported under the TWAIN standard in Win XP. EPSON will only support scanners under the TWAIN standard, therefore it is not possible to install this scanner in WIn XP with EPSON drivers, period. This is a very old scanner and will not have a useful lifespan left, you can buy a more reliable and higher spec unit for less than =A370 ($100).

How about the visioneer 6000 series?

It looks like you're out of luck. After Win 2000 came out, Visioneer said they wouldn't update the 6000b driver to work with Win 2000 - which almost certainly means it won't work with XP either. http://www.ntcompatible.com/story346.html

How are you going to convince me to replace a perfectly working scanner just to upgrade my OS? :) It looks to me like linux is not the only OS that doesn't support all hardware. The difference being, once supported in linux, support never goes away. On the other side, windows support is built by the same people that make the hardware, thus the fast initial support but also the small period of support. When upgrading PC+OS, more likely than not, you _will_ have older peripherals, probably not currently supported by their vendors. So, I'll boldly say that while linux may cost _you_ money, it also may save money to others in other circumstances.

This happens because at least 51 percent of consumers[2] in the United States don't care enough about freedom. Case in point: Michael Badnarik, the Libertarian candidate, lost the 2004 Presidential election.

Well, you could always vote with your wallet and incourage others to do the same... ;-)

Interpretation: buying out any of the companies that can ALREADY read PDFs and convert them to Word is too expensive. Or, interpret as: Microsoft engineering determines they can't do it because they lack the skills and/or knowledge and/or management and/or motivation to actually do it themselves.

FYI: Nuance Communications has already been taken over by its closest competitor, Scansoft. From the looks of this merger, it seems the important factor in people jumping over. I, for one, would rather move to Sunnyvale from Menlo Park, rather than Boston.

I explored offering this service as part of my consulting practice a few years back. Turned out hardware and software weren't the limitations, even if I could use a simple ADF scanner with a reasonably- priced temp staff. The limitation was the time to organize it into usable categories and make it the file name match the type of content, etc. I'm intrigued by some of the new software such as PaperPort http://www.scansoft.com/paperport/standard/ that apparently makes the documents searchable, minimizing the need to title and folder each scanned image, but I'm sure its relying on some kind of OCR and not sure how kind that would be to hand-written docs. As long as it allows a simple interface to tag each scanned image. You're gold.

the audio to text is not there yet. even the best dictation softwares out there are only about 80-90% and leave you to proof the transcript before you give it the thumbs up.
working for a VoIP company I've tested these products, but they do not claim to be 100%. my step father is a director over at chevron and they have many (200+) that use similar software and it does save them time, but bottom line, I don't think Skype would be able to offer him (or anyone) audio to text anytime soon.

Omnipage by Scansoft allows you to scan in paper docs and convert them to XML/HTML. It's not OSS, but it does work quite effectively.

It will also read text-to-speech and support voice control (not sure how well this works). Text to speech requires you open a file in something that allows it to be edited, MSWord, notepad, Dragon's own editor. If it supported a browser it would be a big improvment.

If you go with this, check supported sound cards and microphones on Scansoft's web site. I bought my wife a 6 Dragon USB microphone. Mics are rated 1-6 Dragons and are a key component. There's support for the major sound card chipset, the on board Intel seems behaves.

http://www.scansoft.com/speechpak/talks/ is a text-to-speech software for the blind for use on symbian based (9210 and series60 phones such as 3650, n-gage, 6600 and so on) that does text-to-speech.

anyhow, if you got a friend with s60 phone just give it a spin(you can try it for free). predictive text input is thrown in of course..

i'd imagine a laptop to be quite a bitch to carry around. hell, i'd just skip speaking - would probably be easier... and do stuff like typical phone calls through sms, irc and email.. and just carry around some paper and pencils(provided that he could type fast).

Those of us who must rely on speech recognition software to use our computers not only have no choice but commercial software, we are also relegated to Windows or, with arguably rather less function and accuracy, Mac. There is one open source project of which I'm aware, CMU Sphinx, but its progress is slow and it is not usable as a complete, continuous-speech, large vocabulary speech recognition solution. NaturallySpeaking for Windows is an excellent product and allows those of us with disabilities who cannot use keyboards to do what you see here and many of us who have paid for the software over the years don't begrudge supporting the company but, given its product is not available for any platform but Windows I think this is an excellent opportunity for OSS.

There are others who feel the same. In fact, for a time, IBM had a version of its speech recognition engine available for Linux users but it has since abandoned the project. So, OSS would seem to be the way to go, perhaps building on the very promising CMU Sphinx.

Those of us who must rely on speech recognition software to use our computers not only have no choice but commercial software, we are also relegated to Windows or, with arguably rather less function and accuracy, Mac. There is one open source project of which I'm aware, CMU Sphinx, but its progress is slow and it is not usable as a complete, continuous-speech, large vocabulary speech recognition solution. NaturallySpeaking for Windows is an excellent product and allows those of us with disabilities who cannot use keyboards to do what you see here and many of us who have paid for the software over the years don't begrudge supporting the company but, given its product is not available for any platform but Windows I think this is an excellent opportunity for OSS.

There are others who feel the same. In fact, for a time, IBM had a version of its speech recognition engine available for Linux users but it has since abandoned the project. So, OSS would seem to be the way to go, perhaps building on the very promising CMU Sphinx.

Old and busted = voice recognition

New hotness = word spotting

When are we going to see software for Linux that allow us to search keywords in audio or video files like Dragon MediaIndexer does?

If they want to make it easy, they are going to need to do something like what was done in Kai's PhotoSoap. This is the *only* image editing app (besides iPhoto) that I've seen computer novices be able to figure out. The tools were simple, made sense, and the UI was great. And it worked well, too.

However. How about a Japanese live demo? ScanSoft SpeechWorks Speechify product demos.

Todd

However. How about a Japanese live demo? ScanSoft SpeechWorks Speechify product demos.

Todd

Thanks. I bought a Telex H-851 because it was the only one recommended with 6 stars by Dragon Naturally Speaking, voice recognition software. The Dragon recommendation seemed important, since they want their software to work as well as possible.

/not affiliated with ScanSoft -- just trying to help bring an end to PDF's full of rasterized text

I knew a cop, who about 5-6 years ago, had started using a digital recorder for taking notes -- he had trained Naturally Speaking, and the device had a clean enough recording that when he got back to the office, he'd play back it back to his computer --

A quick review to make sure it had transcribed cleanly, and he was done.

I'm guessing the software's gotten better since then, and the various recorders have gotten more memory, as well. With an iPod, it's not going to be cheap, but it's small, and it'll hold hours of dictation.

It appears IBM already provides this if you buy ViaVoice Advanced:

On the Internet, dynamic navigation of the Web allows Advanced Edition users to access Web sites merely by speaking the link name, while navigating to bookmarks with a simple voice command. Natural Language Commands and Voice Navigation macros provide Command and Control to navigate within the supported operating system and within many popular Windows applications.

Surfing the Web with voice activated linkage, creating e-mail or voice chats are tools within Advanced Edition, which allow users to enjoy the comfort of these productivity features while working on the Web.

So Opera rolled it into their browser. I'm sure that will be less expensive than buying ViaVoice Advanced separately but is this really newsworthy? "Hey, we bought a license to use another company's library!"

Opera's first marketing target: enterprise. Sure, the starship Enterprise had everyone speaking to their computers but in my enterprise I don't want cubicles full of people mumbling to their browsers. I don't see why this would enhance productivity either, except for the disabled. A two button mouse with a scroll wheel will be faster than speaking, certainly.

MS speach recognition is awful. Try Naturally Speaking Not great but from what I've seen and read it's the best out there.

Do nott forget to use a high quality headset (NOT a standalone microphone) such as one from Plantronics* or Logitech and try to get it posistioned the same way every time.

A short history on Naturally Speaking.
It was originally created by DragonSoft. Infact there was an article on the company and the Husband and Wife who started in in MIT's Technology Review back in the good ol days (around 1998) when you'd count off the weeks until the 1 issue per quarter would arrive.

They were then bought out by Lernout and Hauspie who had their own speach recognition software. Dragon Naturally Speaking was then integrated into L&H's product. If I recall correctly the integrated product didn't work very well.

L&H was then bought out by ScanSoft where Dragon Naturally Speaking lives today.

I'm not sure why I bothered to write all this (and I've probably got some of it wrong) this early in the morning. Oh well

-TMF

I've found you can get the plantronics headsets _much_ cheaper online. For example the Audio .90 which sells at circuit city for $30 can be had for $15

MS speach recognition is awful. Try Naturally Speaking Not great but from what I've seen and read it's the best out there.

Do nott forget to use a high quality headset (NOT a standalone microphone) such as one from Plantronics* or Logitech and try to get it posistioned the same way every time.

A short history on Naturally Speaking.
It was originally created by DragonSoft. Infact there was an article on the company and the Husband and Wife who started in in MIT's Technology Review back in the good ol days (around 1998) when you'd count off the weeks until the 1 issue per quarter would arrive.

They were then bought out by Lernout and Hauspie who had their own speach recognition software. Dragon Naturally Speaking was then integrated into L&H's product. If I recall correctly the integrated product didn't work very well.

L&H was then bought out by ScanSoft where Dragon Naturally Speaking lives today.

I'm not sure why I bothered to write all this (and I've probably got some of it wrong) this early in the morning. Oh well

-TMF

I've found you can get the plantronics headsets _much_ cheaper online. For example the Audio .90 which sells at circuit city for $30 can be had for $15

I think you need to look at ScanSoft's product again. Contrary to your statement, Dragon NaturallySpeaking 7 DOES allow hands-free use of the computer. In fact within the past 4 months I have trained 2 people to use it in a hands-free situation. From all of the built in shortcuts for the most popular software (including IE and Office) to the command "mouse grid", which allows you to move the mouse around the screen by only your voice, Dragon NaturallySpeaking will work great. You can even use a command (which I forget off the top of my head) to press specific keys on your keyboard. If he has the patients to sit down and learn all of the commands (which can be done over time...learn mouse grid, then slowly pick up the shortcuts over time), Dragon can be faster than a keyboard/mouse combo. If he has a clear voice, training will take 15 minutes. And it works with XP.

Might I make the suggestion of getting Preferred over Standard. While it's double the cost, it really does seem to do a better job than Standard. Not sure what the difference is (smaller vocab maybe?), but it's there.

SunPin wrote:

Without voice dictation, I can't move to Linux so if you planned to tell me just that, try offering a suggestion instead.

Yeah, dictation on Linux is in a pretty sorry state, and I don't see it getting out of that state without still yet more academic development.

However the latest version of ViaVoice is now available for Mac OS X. That's a serious step up from Windows in stability and system quality.

There's one amazing piece of software: ScanSoft RealSpeak. They have an on-line demo. Lots of languages. It's not perfect, though I haven't heard a better one.

www.expervision.com/webtr6.htm

http://docmorph.nlm.nih.gov/docmorph/ here are some OCR programs

http://www.scansoft.com/omnipage/

http://www.abbyy.com/

more ocr links than you really want http://web3.humboldt1.com/~jiva/ocr/_ocr_resource. htm

You can also get telephone based e-mail from several vendors.

• dyphone-based synthesis where the database contains one dyphone (end of first sound + start of next sound) for each psossible sound combination. This approach is used in Festival. Dyphone-based synthesis will hardly sound better that in Festival because dyphones have to be modified artificially to fit every variation of pitch, duration and any other parameter that is needed to produce a given phrase.
• corpus-based synthesis takes a different approach where a large database of several hours of speech is recorded and manually labelled to mark the start and end of each sound. Such a database is used to extract the best and the longest sequence of dyphones during the production. This approach gives naturally sounding results for short sentences where intonation is not so important Given that the cost of developing a database for corpus synthesis may be orders of magnitude higher than for dyphone synthesis, there are very few companies that make them. Two companies offer a demo on the internet: ATT and Scansoft (former L&H) and

Actually, there is already an equivalent to the proposed Audio tab on Google. It is an HP Research project, SpeechBot

Another similar product is already on the market from Scansoft (formerly Dragon Systems) but it uses a complete different approach than the fast-talk product. We are actually using Scansoft product where I work to index all of the media (audio and video) files on the corprate lan.

It looks like Dragon's "modern" products (Naturally Speaking) are still published (& maintained?) by ScanSoft. Some vendors still show the "obsolete" discrete-recognition Dictate available (Dragon Dictate Power Edition 3.0 $189). If you Google for the part number 01-022-24-01 you'll find other vendors.

L&H also took down Kurzweil (the K. that did voice recognition, not the K. music systems co.).

Kurzweil made a discrete recognition system (Voice) that I used for programming for a while when whateverthefuck is wrong with my hands was kicking up.

I ran it on a Pentium 120MHz laptop w/ 1GB hard drive & it worked pretty well.

I still have the CD's, but it'll only run on Win95, not Win98SE.

Discrete is better for programming because continuous relies on the underlying syntax of the natural language (e.g. English) to hint it at reaonable words for ambiguous sounds. In programming, I say things like "left paren eye spacebar equalsign spacebar..."

Dragon's big advantage over Kurzweil was that it could be used completely hands free (including mouse) so they got the disabled market.

The discrete recognizers had to be trained (Voice called it "enrolling") so they'd know how you pronounce the phonemes, but they had huge vocabularies- you didn't have to speak every word, just enough to represent every sound. The version of Voice I had let me add words, so I could add gss_SetPixelMap & other symbols. There was a way to import a whole list of words, so I munged the output of nm & spent a couple of days training it to recognize every function, type, and variable in the PowerTV operating system.

Kurzweil Voice & Dragon Dictate had extensive "correction" features because they made a lot of mistakes. This started as very irritating, but you'd get used to it & you could talk ahead a few words, then correct the 3rd word back, then continue where you left off. When correcting, it would present a list of close-sounding words to pick from or you could spell out the word & add it if it was new.

There was even a "you keep screwing up" helper where you could give it two words that it confused a lot & re-train them.

One problem with speaking discretely (pausing between each word) is that the back of your throat takes more wear & you can start to go hoarse, causing the recognizer to get worse. Yer supposed to avoid carbonated beverages @ lunch.

Instead of running these old apps on a PDA, why not run 'em on a Crusoe subnotebook or something like the OQO?

• dyphone-based synthesis where the database contains one dyphone (end of first sound + start of next sound) for each psossible sound combination. This approach is used in Festival. Dyphone-based synthesis will hardly sound better that in Festival because dyphones have to be modified artificially to fit every variation of pitch, duration and any other parameter that is needed to produce a given phrase.
• corpus-based synthesis takes a different approach where a large database of several hours of speech is recorded and manually labelled to mark the start and end of each sound. Such a database is used to extract the best and the longest sequence of dyphones during the production. This approach gives naturally sounding results for short sentences where intonation is not so important

XDoc s as others have pointed out is a forms technology, not a competitor with PDF in all areas. However, Adobe purchased Accelio earlier in the year, who make a forms authoring and serving product (formerly known variously as FormFlow, ReachForms, RichForms); Adobe just relaunched the product line a week ago, realigning the company somewhat around server products.

Hence the impact of this announcement. If you've actually used the Accelio stuff (and I have, a lot) you'll know that it could be massively improved upon; other products are biting at their heels already.

So MS weighs in immediately after Adobe's fanfare and says they're going to enter the market (note that XDocs does not even have a release date yet!) - its hardly surprising that Adobe's stock takes a hit.

I first delved into the world of OCR back in 98 with this product and havn't turned back. The current version is made by scansoft (the same makers as OmniPage), this product is much better. Even PCWorld has a review of it (March 2000).

It achieves 98% accuracy on typed text and can handle graphics, bullets and tables. These were big plusses for me. I still use the 98 version and have very few complaints. Dirty pages can be a problem, but it has frequently amazed me in how it catches characters in the midst of goop.

the trick was to use a gray scale scan or text mode, Also I scan at 300 DPI , I find it's important to give the OCR as much info as possible to work from.

I agree. The right settings are very important. I recommend some serious tweaking before you get to hard and heavy with it. For plain text, the above works great, although I sometimes prefer 150dpi. For anything with tables and graphics, 300dpi is a must. A good scanner can make a big difference too. My work uses network ready scanners that copy the file to a network share and the software picks up the files automatically. Very efficient.

I started a small document scanning service a few years ago. (I am no longer in that business). The biggest issue in OCR accuracy is pre-process. (in particular de-skew and grayscale removal). If the page is skewed even a couple of degrees OCR will fail miserably. I have had superb results using TMSSequoia Scanfix software which automatically cleans-up and straightens the page nicely. Its expensive but worth-it if you have a lot to scan. I believe that they still have a demo available.

My experience has been that the consumer OCR software is considerably MORE accurate than industrial versions that cost 20X as much. I obtained excellent OCR accuracy using Scansoft's Textbridge software which utilized the Xerox Textbridge engine. Scansoft appears to have purchased Omnipage OCR and discontinued the Textbridge OCR line. I found that I achieved much higher accuracy with Textbridge then with Omnipage after the document was processed by Scanfix. Textbridge did not have some of the features of Omnipage but Textbridge was faster and better at OCR. I would definately download the Textbridge 98 demo that is still floating around on the web.

Both Textbridge and Omnipage OCR were vastly superior to anything else I previewed, including Adobe's OCR engine. OCR can be surprisingly accurate but the source image needs to be free of distortion. Sometimes you will need to break up the page into several using photo-editing software since no OCR can inteterpret the structure of a document very well.

I suspect that you will be better off just typing in the mathematics in by hand. Maybe a visual LATEX editor like Scientific Workplace would be helpful. The LATEX output could be manipulated using a parser to put the equations into the simpler forms that you need while keeping the raw equation in a form that could be used for other purposes later on.

Honesty, 10pgs/hour is pretty good so it doesn't sound like you are doing all that much touch-up. I suspect that using Scanfix will provide the greatest boost in productivity.

I started a small document scanning service a few years ago. (I am no longer in that business). The biggest issue in OCR accuracy is pre-process. (in particular de-skew and grayscale removal). If the page is skewed even a couple of degrees OCR will fail miserably. I have had superb results using TMSSequoia Scanfix software which automatically cleans-up and straightens the page nicely. Its expensive but worth-it if you have a lot to scan. I believe that they still have a demo available.

My experience has been that the consumer OCR software is considerably MORE accurate than industrial versions that cost 20X as much. I obtained excellent OCR accuracy using Scansoft's Textbridge software which utilized the Xerox Textbridge engine. Scansoft appears to have purchased Omnipage OCR and discontinued the Textbridge OCR line. I found that I achieved much higher accuracy with Textbridge then with Omnipage after the document was processed by Scanfix. Textbridge did not have some of the features of Omnipage but Textbridge was faster and better at OCR. I would definately download the Textbridge 98 demo that is still floating around on the web.

Both Textbridge and Omnipage OCR were vastly superior to anything else I previewed, including Adobe's OCR engine. OCR can be surprisingly accurate but the source image needs to be free of distortion. Sometimes you will need to break up the page into several using photo-editing software since no OCR can inteterpret the structure of a document very well.

I suspect that you will be better off just typing in the mathematics in by hand. Maybe a visual LATEX editor like Scientific Workplace would be helpful. The LATEX output could be manipulated using a parser to put the equations into the simpler forms that you need while keeping the raw equation in a form that could be used for other purposes later on.

Honesty, 10pgs/hour is pretty good so it doesn't sound like you are doing all that much touch-up. I suspect that using Scanfix will provide the greatest boost in productivity.

Our product was aptly named TFS Translator. The engine was called Barcelona, made by Globalink, until recently Lernout & Hauspie and currently ScanSoft. (Some people have too much money) Link to the current product based on that engine. They also have some tips on what to keep in mind if you are preparing a text for automatic translation. Ah, they finally added Japanese! Konnichi-wa, Anjin-San! :-)

...and a balrog in a pear tree!

Scan to OCR to PS to PDF

there are apprarently a couple tools to do this for you. check out a brief list here

Seeing as you've looked into Adobe Capture, windows may be an option. If so, then the other question would be whether you've looked into Textbridge? This looks like it would do exactly what you're asking. No muss, little fuss.

As I understand it, Goo or Soap are NOT part of this particular deal. Read carefully -- neither is mentioned in this article. Why? Because Goo and Soap are owned by ScanSoft, a company that has quietly been acquiring quite a number of products. Will they pay any attention to the Linux market? I doubt it. They're more likely to pay attention to how their products can be used over the Web.