Olympus WS-400DNS Hands-On Review

We received a lot of emails about the new Olympus WS-400DNS with Dragon NaturallySpeaking v10 Recorder Edition since it was announced earlier this summer.  Why? It's the most accurate voice recognition product for PCs in the world - normally selling for 100's of dollars as a stand-alone application that translates speech to text and execute commands in real-time. It has never been available to consumers at such a low cost, less than $150 bundled with a pocket voice recorder.

What's the difference between this version, and the full version of Dragon NaturallySpeaking?
Real-time voice transcription and voice commands are disabled in the Recorder Edition, you'll have to upgrade for those features. The Recorder Edition only allows transcription of recorded files, from a recorder or another piece of software.

Out of the Box

We wanted to see how easily we could get Dragon installed, configured and begin transcribing our recordings from the WS-400 recorder without even looking at a manual.

What's included?

• WS-400S Digital Voice Recorder
• ME-12 Noise Cancelling Microphone
• Dragon NaturallySpeaking DVD
• DNS Recorder Edition Software License
• Carrying Case
• AAA Alkaline Battery x 1
• Instruction Manual
• Warranty Card

As you can see the recorder is is quite small. Small enough to fit in a shirt pocket and forget its even there. It uses one AAA battery, so its very light, even with the detachable microphone plugged in.

The buttons are small but arranged in such a way that your thumb can feel its way around the unit without having to look... with a little practice of course. Each button has its own shape and feel: the record button is on the right, the stop button on the opposite side has an obvious bump designed to help you feel your way around.

We were suprised how quickly we got used to navigating the menu, browsing the file folders and of course, setting the time and date without even looking at the manual. It was that easy.

Using the WS-400 Recorder

How does it work? Click on the record button to begin recording, a numbered file is automatically created and named "ws400001.wma". Click Record again to pause your recording. A third time continues the recording where you left off. The Stop button completes the current recording and any new recordings will be written to a new file numbered sequentially

Use the up/down scroll buttons to navigate the list of recordings and click play (center button) to play it back.

Our first "check, one, two three.." recordings were recorded with the detachable microphone. The playback however was barely loud enough. There is no dedicated volume wheel on the side as you would expect. Though with trial and error, we discovered the up/down navigation buttons double as volume controls when playing recordings.

The sound quality on playback was a bit thin through the tiny speaker on the front panel. There was also noticable noises in the recording-windy plosivies and sibilance from both breath and pronouncing T's and S sounds. A human can hear past these extra noises but will the voice recognition handle them? Can Dragon transcribe accurately with all the extra noise? We'll find out.

There is no obvious on/off button. It does turn itself off when not in use for 5 or so minutes. Later (after looking at the manual), we learned that by holding the "Hold" switch in the UP position for a few seconds, it turns the unit off. This is a good feature, it eliminates any accidental power downs from misplaced button clicks by requiring you to move a switch instead. If you push the switch up by mistake, you have a few seconds to push it back without it powering down.

Setting up Dragon NaturallySpeaking

Installing Dragon NaturallySpeaking was rather straight forward and took about 30 minutes, including the 15 minute voice training. The software requires you to type in the license key code during installation... so don't lose your license code!

We installed DNS on a Vista Business 32bit, 1.8ghz Dual Core PC with 4Gb of RAM. When the installation was complete, the "New User" wizard is automatically ran. This part can be skipped if you're short on time but its necessary to begin training DNS to your voice.

What about my accent?

Training DNS to your voice is necessary to improve its speech recognition accuracy. If you're worried about your accent, Dragon supports a number of foreign and regional accents out of the box.

Built-in Accent Profiles

• General—Use this if your accent is not covered by the other choices or you are not sure which accent to select.
• Australian accented English
• Indian accented English.
• SEAsian accented English—South East Asian accent
• Inland Northern US (Great Lakes area)—This accent model covers Upstate New York through the Chicago area.
• British accented English.
• Southern US—Accent model that covers most of the southern United States, including Texas.
• Spanish accented English.

I was born and raised in the South but my accent is mixed. So I selected the General accent profile, which is already selected by default.

After selecting your accent profile, the New User wizard displays a half dozen excerpts from a number of books. The excerpts are quite lengthy but reading all of them isn't necessary. You're only required to record 15 minutes worth of speech to get started. Of course, the more you record, the better the voice training and the transcription accuracy.

All you need to do is read one of them for at least 15 minutes while recording your voice on the WS-400 recorder. We initially recorded this training with the detachable microphone. The recorder displays the length of the recording as you go along. So keeping track of the time is not an issue.

When reading the book excerpt, we paid close attention to how we held the recorder. We wanted to make sure we were holding the microphone directly in front and minimize any unnecessary sounds and words.

When the recording was complete. We connected the WS-400 to the PC using its built in USB jack. The jack slides out the bottom of the recorder using a slide switch on the left side. Our PC has a USB jack on the front panel making the connection easy and convenient. However, if your USB connectors are in the rear, we'd recommend getting a USB extension cable or a USB hub.

Once the voice recorder was attached to the PC, the PC recognized the device as a removable drive and automatically mapped the voice files to drive letter F.

Back to the New User Wizard, select the file containing the 15 minute voice recording of the book excerpt you read. Dragon will analyze the recording taking about 12 minutes total on our PC.

Note: Each time Dragon transcribes your recordings, it learns more about your voice... the more your record, the better it gets.

After the New User Wizard completes its analysis, it automatically types out the recording in DragonPad, a Microsoft WordPad like text editor.

After the recording is typed, you have the option of having it play the recording while highlighting the words at the same time. The recording and highlighted text are working in perfect sync. This is quiet handy. Playing back and highlighting your recordings at the same time allows audit the transcription and correct mistakes as you go along.

How well did it do?

Suprisingly, the transcription was perfect. Every word, even the big ones were transcribed, like "interoperability" which has 8 syllables. I found myself slowing down when pronouncing some words because I wanted to make sure it was understood as best as possible. But most of the time, I read at a normal speed.

What DNS didn't do however, was punctuate and paragraph the transcription. The entire transcription was typed out in one giant block of text. I would have to go over the typed document and make the corrections manually.

Lets try again. I remembered seeing a demonstration of voice to text on television where one had to actually speak the words, "period", "comma", hyphen", "next paragraph" etc. while dictating into the recorder.

So I began a new recording by reading an online news article, only this time, speaking out the commas, periods and paragraphs as I went along.

Once done, I open the DragonBar, clicked on "Transcribe Recording", click "Browse..." to the newly recorded file and watch it type it out in real time in DragonPad.

To our amazement, Dragon understood my instructions and formed the paragraphs and sentences perfectly, exactly as I spoke them... even the hyphenated words.

Remember, we've still not looked at the manual at this point

Was the typed document perfect? No, but it was close. It didn't handle the capitalization of proper nouns, for example, "Joe Schmoe's Bar and Grill". These would have to be corrected and added to the vocabulary options in Dragon as you discover them. However it did handle acronyms like "PST", as in "Microsoft Outlook .pst files". I never pronounced the "dot pst", just "pst". Nonetheless Dragon didn't interpret the acronym as a filename extension, instead capitalizing the letters as is should do for 99% of  all acroynyms.

But still, we carefully compared what was read to what was typed and we were amazed at how accurate the transcription was.

Does Dragon support Outlook, Microsoft Word, WordPerfect?

Yes to all three. When you select a file for transcription, you have the option to insert the text into DragonPad or by selecting a Window (eg. Outlook or Word).

Final Thoughts

We expected DNS to have a lavishly complex user interface with 100's of buttons and hidden dialog boxes, but that wasn't the case. Despite its sophistication, the application simply consists of the DragonBar, a tiny tray application you click on to import your newly recorded voice files.

Some of our customers have called asking questions expecting the software to do everything: Upload the recording of anyone's voice, transcribe it perfectly with proper form and paragraph spacing, automatically print and email the document to their boss. Sorry, Dragon doesn't do all of that, especially the "Recorder Edition". But what it does do, it does it very well. Considering most people speak over 120 words per minute, but type less than 40 words a minute. Dragon NaturallySpeaking lets you create letters and e-mails about three times faster than typing by hand!

But for those who have a large volume of dictations to transcribe, and are willing to learn the application and use the correction features to further enhance its vocabulary and accuracy, then this product will benefit you greatly, especially considering its extremely low cost at $149.99. Of course, if you need more than what this edition gives you, you can upgrade to the Preferred Edition and save money.

Product Description

On-the-go digital audio recording with on-the-mark voice activation built in and industry-leading speech-to-text software included. Let this sleek, stylish recorder go the distance for you by providing 272 hours of superior stereo sound recording, with features that maximize your ability to organize files and transfer them to your PC—without even using a USB cable. Comes with a state-of-the-art noise cancellation microphone, too! Recording and downloading crisp digital files has never been easier.

Advanced ‘Dragon Naturally Speaking’ speech recognition software from Nuance, included with the recorder, lets you convert dictation, notes, lectures and other audio directly into text, even while you’re on the move.

This digital recorder has 1GB of internal memory, which translates to 272 hours of recording time, enabling you to store as many notes, lectures, interviews and conversations as you need to—with extra room for documents.

Quick and easy! Just plug the recorder right into your PC to transfer all files. This USB Direct PC link eliminates the need for cables and extra work to get the job done.

That would be WMA—Windows Media Audio—which is one of the most popular and functional recording formats in the world of portable digital recording.

The unit also comes with our state-of-the-art ME-12 microphone, noted for its tremendous ability to capture high-quality audio with no extraneous sounds.

Since the WS-400S DNS can be set to record only when the microphone senses sound, recording time and memory can effectively be extended.

Key Features

• Sophisticated speech recognition software, included with the recorder, enables on-the-go audio-to-text transcription.
• Ready and set to go—and go and go—with 1GB of internal flash memory and 272 hours of high-quality recording.
• USB Direct Link for cable-free PC downloading of all files.
• Records in the industry-standard WMA (Windows Media Audio) format.
• A technically-savvy noise-cancellation microphone comes standard with the WS-400S DNS.
• Voice Activation makes automatic starting and stopping of the recorder an extra benefit that saves time and energy.
• With five built-in folders, each holding 200 files, quick retrieval and storage is a given.
• The bright LCD supports and displays three languages—another plus for accessibility.
• Long battery life—21 hours of continuous recording—gives you the on-the-go confidence you need (Alkaline, LP mode recording).
• A wealth of features and software—either bundled or built right in—all in one practical, convenient package, includes speech-to-text software, high-function microphone, 1GB of memory, USB Direct Link and much more.

Additional Files

• Check back later

Package contents

• WS-400S Digital Voice Recorder
• ME-12 Noise Cancelling Microphone
• Dragon Naturally Speaking Ver.10 Recorder Edition Software & License Code
• Carrying Case
• AAA Alkaline Battery x 1
• Instruction Manual
• Warranty Card

Technical Specifications