simple digital recorder question Options

[mil]

Hi avid knowledgeble team members,

I don't understand the complexity about digital recorders. Can you please help me with the following.

1. Can DNS6 use ADPCM audio files? I want to buy an olympus VN 1800 and it saves in this mode.

2.The help menu tells me that it can transcribe .wav files. Is ADPCM a wav file?

3. What files formats does DNS work seemless with?

4.Can't I use any digital recorder, and convert it to the acceptible DNS format with a third party program?

5. Are there is other issues I should know when buying a recorder? Is there a general of thumb.

Kind regards,
Mil.

[G_Tom S_*]

Hi.

For a recorder to work with Voice Recognition, it's important that it be able to transfer audio files to the computer via a USB connection. In addition to the benefit of very rapid transfer of audio files, USB is good insurance that the volume-level of the audio recordings will stay at its most optimum level.

Somewhat similar to old-fashioned tape recorders, the $60 Olympus VN1800 doesn't have a USB connection. Its only way of transferring an audio recording to your computer is to connect the recorder's EARPHONE jack to your sound card's LINE IN jack, and then record that sound with a program capable of creating a WAV file.

The audio volume level from the recorder is fully dependent upon how you set the recorder's VOLUME control during each transfer. That's too much of a variable, because of somewhat strict sound volume-level requirements when setting up and training a new user with recordings from that recorder, and also when later feeding all of your WAV files into DNS for voice recognition.

Olympus recorders with a USB connection start in the $200 range, such as the Olympus DM-1 model. I've had one for several months now, and its overall capabilities are pretty awesome. Highly recommended. I'm currently experimenting with various headset mikes with the DM-1, but even the built-in mike regularly gets better than 95% VR accuracy with Dragon Naturally Speaking.

Regarding your questions:

>>>Can DNS6 use ADPCM audio files? I want to buy an olympus VN 1800 and it saves in this mode.<<<

---ADPCM (a method of file compression, to save RAM) is only used internally in the VN1800. Therefore DNS6 will never see ADPCM so it doesn't matter.


>>>The help menu tells me that it can transcribe .wav files. Is ADPCM a wav file?<<<

---No, but once again, it doesn't matter. For a digital recorder, the VN1800 has quite limited features. There's no way for a digital ADPCM file to leave the VN1800. Only analog audio, via an EARPHONE jack, will get out.


>>>What file formats does DNS work seamless with?<<<

---WAV files. Working example: When an Olympus DM-1 internally records and saves audio files, it compresses the audio into Olympus' special DSS format and then stores that highly-compressed DSS file on a removable 64-MB memory stick. (With the DM-1, a 64-MB stick holds over 10 hours of nonstop best-quality recording) When it's time to upload your audio files to a computer, you connect a USB cable from the recorder to your computer (using WinXP, in this example) and the computer immediately recognizes the recorder, treating it just like another disk drive.

The Olympus computer program that handles DM-1 files can be set to automatically download the DSS files from the recorder, un-compress them into WAV format, and then save them on the computer in a default folder of your choice, non-stop. This goes quite rapidly, for example with a 1.6 GHz machine a one-hour-long DSS file downloads to your computer in approx. 4 seconds, and then un-compresses to WAV format in another 10 seconds or so. It's now ready for Dragon.

Disconnect the DM-1, start Dragon, activate the "Digital Recorder" user, select SOUND, select TRANSCRIBE RECORDING, and a window pops up showing all of the WAV files in the above-mentioned default folder. Select the WAV file of your choice and off it goes, typing like mad across your screen. It's very fast and seamless, actually.

Note of interest: Similar to MP3 audio compression, a typical DSS compressed audio file is just one-tenth as large as the same identical audio data in WAV format. Thus DSS files save a lot of space/time while storing and emailing, for example.


>>>Can't I use any digital recorder, and convert it to the acceptible DNS format with a third party program?<<<

See above. However, for excellent VR accuracy with a digital recorder, DNS would prefer rock-solid predictability regarding your voice being clearly enunciated at the same volume level, in a fairly noise-free environment. Exactly how you accomplish this, is up to your expertise. Sometimes it seems easy and other times it takes a concatenation of miracles, horse-sense, and dancing around the campfire counterclockwise three times in succession.....

I prefer CoolEdit 2000 as my third party audio program of choice. It has a lot of conversions built in.


>>>Are there other issues I should know when buying a recorder?

Be ready to spend a minimum of $400--$500 for a decent recorder plus Dragon plus headset mike for the computer. This VR technology is truly amazing, so don' t sell it short by scrimping on equipment and then being disappointed because it didn't work the way we say that it does.

Regards,

Tom Swanson

[mil]

Tom,

Thanks for your informative reply. I appreciate the time it takes to reply so thoroughly!

One more question I have as budget is an issue for me. Olympus also have a DW 90 model which as usb connectivity - this means you can download adpcm wav file.

Would this be suffice? Would other voice recorders with usb wav capabilities be suffice also. Or shouldn't I even waste my time with them?

Your response would be very must appreciated.

Kind regards,
Mil.

[G_Tom S_*]

Hi Mil,

I downloaded the DW-90 user's manual and it says that the DW-90 uploads WAV files to the computer, so that would make it workable with Dragon in that respect.

I remember looking at the DW-90 last summer before buying my DM-1, and the thing that made me back off from it was the fact that it will only record about 30 minutes worth of VR-quality sound in the Standard Play (SP) mode, and there's no way to pop out a memory card and thus upgrade its internal storage memory, so you're stuck with that limitation.

Sound recorded in the Long Play (LP) mode won't work with Dragon because Dragon simply refuses to load a WAV file that's digitally sampled at such a low repetition rate as the LP mode offers.

--So if you can get by with just half an hour of recording, then yes, it will work.

HOWEVER: I see where Olympus DM-1 prices are going down, with current street prices in the $170 range, while I paid $230 for mine just 2-3 months ago.

Honestly, I recommend that you do whatever you have to do, to get this model instead. Take a look at these reviews:
http://view.atdmt.com/AVE/iview/bzrtcaws10...hi.60/01?c****=

(Prediction: This forum's automatic word-censor might erroneously put a string of ****'s on the last word of that web address because it mistakes "c l i c k" for "l i c k", so you might have to manually type c l i c k with no spaces, followed by an equals-sign, in order to pull up that website with your browser.)

If there's any way for you to make an income with the Olympus DM-1 and Dragon so that it will pay for itself, then beg, borrow or steal to get it, because recorders at the level of the DM-1 truly represent a handheld-recorder breakthrough of significant magnitude.

Plus, I never thought that I'd have use for the DM-1's MP3 music-player feature, but I recently used the included software to translate to MP3 format and then copy a full hours' worth of CD music (Ladysmith Black Mambasa, in this case) directly onto the DM-1 and instantly fell in love with the hi-fi quality of sound that this tiny little unit is capable of producing. I agree with the multiple 5-star reviews; it's quite stunning.

--And it works great with VR too!!!!

Regards,

Tom Swanson

[mil]

Thanks again Tom, your feedback is going to save me a great deal of frustration and time.

The DM1 model is selling an Australia for over $900 Aust - so you can see my delima in find a cheaper alternative.

A quick question on how effective the recorder will be. If I do buy the DM1, will the recorder work if I sit in different environments such as a car? If you can give me a brief description on how you use it and how effective it would help my decision to spend the big bucks.

Also, ebay is selling some units that claim to be powerful and of great quality (32000 k sampling).
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewI...&category=14960

It's title is 44 digital voice recorder.

The less memory model is sellingfor around $140 - the specs suggest it is high quality.
My final question (and for give me if I a am stretching the friendship)...is this worth considering or would you stick to olympus still.

Once again thank you so,so much. I hope you are using DNS to input these responses (as I will not feel so guilty).

Kindest regards,
Mil.

[G_Tom S_*]

Hi again, Mil.

>>>The DM1 model is selling an Australia for over $900 Aust - so you can see my delima in find a cheaper alternative.<<<

Wow. That's over $400 US. You should somehow be able to cut that price in half, at least.

>>>A quick question on how effective the recorder will be. If I do buy the DM1, will the recorder work if I sit in different environments such as a car?<<<

Yes it will. Just make it a habit of dictating in the same fashion wherever you are. Find what distance from your mouth the mike works best for optimum VR, and keep that the same as much as you can. Practice enunciating each word clearly and distinctly, because clearness of dictation is more touchy with a portable recorder than it is with a headset mike on a desktop computer.

NEAT CORRECTION FEATURE: Remember, DNS (Preferred and above) keeps a temporary audio record in addition to the VR that you see on the screen. Therefore when there IS an error, all you need to do is highlight the questionable word/phrase with the mouse and give the "Play That Back" command, and you'll hear the exact words that you spoke into the recorder. This makes error correction so easy, that even using a digital recorder in a noisy background, you can achieve almost-guaranteed 100% accurate duplication of what you said, just by doing the "Play That Back" correction step. When there are more errors it just takes a bit longer to proofread, is all.

>>>If you can give me a brief description on how you use it and how effective it would help my decision to spend the big bucks.<<<

I bought the DM-1 a couple of months ago with the intention of just using it for various note-taking tasks, which it is really good at. For example, my daughter is a geology major who takes extended summer backpacking field trips with her classmates. All she needs for excellent detailed geology field notes on a 2-week camping trip is her GPS, digital camera, and digital recorder, and all three together barely weigh one pound.

In early November I found that the DM-1's features made it a natural to use with VR, and I bought IBM's ViaVoice10 Pro in hopes of doing so. ViaVoice repeatedly crashed my XP system enough times that I finally discarded it and bought Dragon 6 Preferred, which hasn't had a fraction of the system problems that the IBM product did.

I experimented with the DM-1 and Dragon, and the accuracy was so surprising (typically better than 95%) that it made me sit up and take notice. I'm still in the experimental stage -- I'm semi-retired in the winter, meaning this is my "hobby-time" -- so whatever I do with it just now can only be classified as hobbying.

For some professional opinions, check out this website: http://www.dbic.net/Prod9-RC.htm

That site speaks highly of the Olympus DS-3000, a close cousin to the DM-1, but more expensive. Note that the DS-3000 on that website claims just 2.5 hours of VR-class recording with a 16 MB memory card, but that must be old news because the DS-3000 (and DM-1) today come with a 64-MB memory card, so they both get 10 hours of VR-class recording right out of the box. You can also upgrade with a 128-MB card if more storage is needed.

>>>Also, ebay is selling some units that claim to be powerful and of great quality (32000 k sampling). http://cgi.ebay.com/ws/eBayISAPI.dll?ViewI...&category=14960 <<<

That unit has almost the same features of the DM-1; I wonder if the same company made the electronics? However, if you do the math, then their claims seem to stretch the imagination, because with a 32K sampling rate there's no way you'd get 10 hours of recording on a 64-MB memory card.

The DM-1 and DS-3000 both use a sampling rate of 11K and compress the audio at a 10-1 rate, which is about the limit, and still only get 10 hours on a 64-MB card, so I think that ad is a bit misleading. They're comparing apples and oranges.

The main reason that ebay unit is questionable to me, is that it doesn't have a name brand! Let's say it goes sour in a couple of months; who stands behind it? I bought my DM-1 at a local CompUSA store along with their 3-year extended warranty, so I'm confident that it'll be working as advertised for at least the next 3 years, without gambling. I try to get the extended warranty on any hardware that I buy, just for that reason.

>>>I hope you are using DNS to input these responses (as I will not feel so guilty).<<<

:-) Crank up the guilt, Mil! I alternate between using DNS and just plain typing, because I also enjoy typing. No problem; please understand that your particular questions are also a valid representative of quite a few other people who browse thru this forum, and so if I and others on this forum are able to give comprehensive and accurate guidelines/answers here, then EVERYONE with your interests and concerns can make use of that info, and thus we won't need to "re-invent the wheel" and repeat it over and over. This discourse has a certain "amplifier" effect, in other words.

Thank YOU for the pertinent questions, sir!

Regards,

Tom Swanson

[Griz]

Really interesting stuff because I have used DNS for a few years now - upgraded to v6 legal about a year ago. However, still using the old Dragon Recorder sporadically because of only 80-85% rec accuracy.

Purchased an Olympus DS 3000 based on great recommendations almost a year ago. Unfortunately it sits in my desk drawer. Why? Because the accuracy was AWFUL - 50% at best. Could not get it to work well! Spent a LOT of time training it and trying different things (different USB ports, in a quiet room, in a car, with/without a headset, Win 98SE and W2k, with/without *********** ect..., and yes, made sure to use the SP rather than EP config) I can get it to transcribe to .wav files in it's own software no problem, and the sound plays back wonderfully both thru the recorder itself and the .wav files on the PC. DNS will accept and transcribe the .wav files but will not or cannot transcribe them accurately. The OL retail vendor was sympathetic but not helpful as DNS is not the endorsed VR vendor (ViaVoice is) and wanted a 15% restocking charge. A call to Olympus tech support was amazing (in a bad way). They advised they do not directly support the product(!) rather support is farmed out to trained retailers. Strange but true. I'de love to have it work as per the rave reviews but after so many hours over so many days it became an unrecoverable cost.

Aside from that experience - love DNS. If anyone's got any suggestions on the DS 3000 issue I'de love to hear them before I get around to dumping the recorder.

[G_Tom S_*]

Hi Griz,

Wow, I wonder what's wrong with that setup. Do a Google search for "DS-3000 accuracy" and you'll find lots of good press on that recorder. For an example of the VR results of my DM-1 (almost identical to the DS-3000) check this out:
http://www.voicerecognition.com/forums/ubb...TML/000822.html

Say, why don't you email me a DSS or WAV file of a short recording (under 200K in size) made by your DS-3000? "Testing 12345, this is a test of the Olympus DS-3000 recorder" will do just fine. If the recordings are as clear as you say, then it's not the recorder's fault, which is what you said in your post.....

Hmmm. Could be the Dragon input level has gone doofus somehow. Just how did you train the Olympus user file, anyway? --How MUCH training, that is?

Regards,

Tom Swanson
tswans@gci.net

[James]

We currently use the DM-1 for the dictations with the doctors. What we have found is that often the doctors will talk about many different patients and at the end of the day will give the recorder, some notes are short, some are longer. We used to have a Sony recorder but an Patient actually stole it from the office. However, the Sony software is REALLY good for porting the .wav files into dragon. So basically we take the recorder, d/l the dss files, convert them to .wav for processing, use the sony software which quickly assembles and puts them into dragon, then we take the dss wave files and place them on the server for transcription (proofing/processing of the VR files) This setup works well, but it would be nice to find a way to have one peice of software to process the files. I hate the VR recording after dragon does it's thing, because if you edit the text file, the recording gets shafted. We're at about 90% accuracy with the dictations using this method and the DM-1. Any ideas on how to simplify the processing?

[John Wickett]

James,

Sometimes the "tried and true" method is still the best approach. Let the doctors dictate all they want into a recorder and the have someone transcribe the dictation using DNS. Yes, they are re-dictating the text, but the increased recognition makes this method the most cost effective at this point in time.

John

[Griz]

Tom S: Thanks for the reply. I purchased the 3000 after reading several reviews - that's what's so puzzling. Over an hour of training which should be sufficient to bring better accuracy than 50% - I'de train more but after the results - no time. (Plus, with training the DNS using the headset for in office use, I cannot read the DNS story lines anymore without falling asleep!)Trained it by creating a new mobile user with USB connection specified. 1 story read in a relatively quiet office enviroment, and one outside with higher, but not excessive, ambient noise. I file protected my dictated training stories on the 3000 to reload after making some adjustments (more RAM, new soundcard), still no luck. Tried all 4 usb ports. System is w2k professional, 550 processor, 512k RAM, creative soundblaster, Version 6.1 legal. So I tried it on my secretary's machine (different newer hardware, same O/S, same software) - same results. I haven't had any of the other lawyers try it as they're so busy - and I'm the guinea pig anyway. When I get a chance in a few days or so and I'm back in the office I'll email you a file or two for analysis. I really appreciate that. My regular in office use of DNS is really good - with a handheld ASM-1 microphone to boot! Can't stand headsets. Used/trained it so much I can hold the mic about 12" away and still have great accuracy. Never did a real calculation per your post of 1/31 but my guess is 95%+. The dragon recorder isn't as good but far better than the 3000 right now. Sidebar - also talked to ***** one day who apparently has had pretty good success transcribing the wav files with regular headset trained user profile. Go figure.Only other thing that's an unknown is the Pro Player software. In loading it on different machines it would be a little finicky - asking for the license code a couple times and getting hung up, but it did load. I also downloaded the latest patch for the software (Olympus doesn't tell you what bugs it fixes) without any difference. I don't have any other software to try. P.S. I emailed Olympus support a couple days ago asking about any bugs/problems feedback and briefly explained my situation. I received a reply today - soft of. They emailed me a .pdf version of the 3000 manual.

I'm overwhelmed.

James: Sounds like you need a macro guru to get your process streamlined. That's way way beyond my skill level.

John Wickett: Yep, transciptionist or secretary is still most technically accurate - but I'de rather have them doing other things. Especially if I have to compose long or complicated documents. It's a lot faster for me to proof my own shortly after dictating rather than having to trade a rough draft back and forth. Noone in my office has figured out how my thought patterns work yet (never will either!). It's a trade off.

Thanks folks.

[G_Tom S_*]

Griz,

This is an interesting challenge that I think we can overcome, if you have a high-speed internet cable connection. This method is used by reporters with overseas assignments, so it should be a no-brainer to do.

1) Email me (tswans@gci.net) a couple of your compressed DSS Dragon training files. Don't convert them to WAV format; I'll do that on my end. Using those, I'll set up a new DS3000 Dragon user on my system, using your voice files.

2) Also send me one or two sample 5-minute DSS voice recordings of any material of your choice, and I'll convert them to WAV format and run them thru the new Dragon user, and see what happens.

This should work pretty slick. I can also use CoolEdit to double-check the audio signal level, signal-to-noise difference, dynamic audio range, and any introduced distortion etc. in case of any perceived trouble. --Matter of fact, I'll probably take a look at all of the files with CoolEdit before bothering Dragon with them. Any problems will be evident up-front, that-a-way.

Let's give it a try.

Tom Swanson

[Griz]

Outstanding. I'm gonna shoot for Monday AM - don't hold me to it though, I'm gonna walk into a circus Monday, but I'll try to get them out to you before it gets real craxy. If not Monday - by the end of the week. Many thanks.

[G_Tom S_*]

Looking forward to it.

Tom

[Griz]

Tom:

You are not going to believe this, however, I was preparing the training and sample files to send to you and just for the fun of it I installed and reinstalled the Olympus software along with the update. (mind you I've done this before) I am not sure what this has to do with anything, however it appears that Dragon is now recognizing my speech-both as a trained Olympus user and a regular PC user. Lo and behold, it actually seems to work. However, I do not quite trust it yet so I am keeping my DSS speech files just in case. If I run into further problems, I will not hesitate to contact you. Thank you very much for your offer.