IPB

Welcome Guest ( Log In | Register )


 
Reply to this topicStart new topic
> DNS (or other software) for meeting minutes?
arcuhtek
post Sep 9 2008, 02:53 PM
Post #1


Newbie
*

Group: Members
Posts: 1
Joined: 9-September 08
Member No.: 21,914
Which Program?: DNS
Why do you Use VR?: work/architect/meeting minutes
Important Urls in your life: www.mastersgentry.com
Where are you Located: Asheville, NC



Hello all!

I am new to VR software (and digital recording) so please bear with me. My question is simple, so I will ask first then give you some background.

I want to use a digital recorder (Olympus Ws-311M) to record meeting minutes in a conference room with between 4 and 10 people, then be able to transfer the resulting file to my PC and use a VR software to have the meeting converted into a Microsoft Word file that I can polish up and distribute to attendees. Is this possible with any version of DNS or other VR software?

Background.

Before you say search, I have already done so and read a few threads, which seemed to direct each poster to other threads about learning how to speak correctly and clearly or basically train their speech patterns such that the spoken statements interfaced most optimally with the software (DNS).

That is not possible for me, as per what I am suggesting, because anyone in the meeting will be speaking (I am sure some speech will be jumbled as people talk over one another), and I cannot reasonably expect everyone to speak with a "trained voice" for my own purposes.

I already have the recorder purchased and have recorded one 2 hour meeting and successfully transfered it to my PC. It is awaiting my next step of purchasing software to do the automatic translation OR......I can do it the old fashion way - put my ear phones on and manually translate it which will take me DAYS to do at my slow 40-50 wpm typing rate.

My expectation of a VR software is that it understand and translate 75% of the spoken words correctly and I do not even care about formatting. I fully expect to have to correct some words, insert a persons name beside each of their spoken statements and I am really fast at MS Word in terms of software efficiency, formatting, distribution, etc. (just not a fast "typist").

I would love to be able to use this method of taking meeting minutes often.

Please know I am only looking to anyone on this forum to give me a general idea if DNS (or other) software is purchased, that it can very basically covert meeting minutes into some form of digital text. I am not looking for a miracle. Does anyone recommend a free trial of a software that might accomplish my tasks? I hate to spend a lot of money just to try it out on one file.

Thanks!!
Go to the top of the page
 
+Quote Post
KnowBrainer Tech...
post Sep 9 2008, 06:41 PM
Post #2


Advanced Member
***

Group: Members
Posts: 1,094
Joined: 9-July 06
From: Nashville, Tennessee
Member No.: 19,645
VR User?: 20 years
Which Program?: Dragon Medcal Practice 11 & KnowBrainer 2011
Why do you Use VR?: We're 1 of the World's Largest Nuance Certified NaturallySpeaking Resellers
Where are you Located: Bethpage TN - USA



Not all digital recorders are created equal. Some are better than others but none are going to be able to do what you want to do. It seems about once a week someone asks this question on the KnowBrainer Digital Recorder Forum but the answer is always the same. What you want to do isn't even remotely possible and here's why:

NaturallySpeaking is only designed to transcribe recordings from a single individual speaking in close proximity of a digital recorder microphone. The recording also needs to be transcribed into a proper digital recording user profile that was previously created by the person making the digital recording. What you are trying to do is literally 3 steps away from reality. Sorry

It really doesn't matter what digital recorder you choose to use because what you want to do simply isn't possible. However, you can Echo Dictate the minutes into NaturallySpeaking over your own microphone user profile.


--------------------
Lunis Orcutt - www.knowbrainer.com
Chat Live w/a Nuance Gold Certified Solutions Provider
Go to the top of the page
 
+Quote Post
Chucker
post Sep 9 2008, 08:29 PM
Post #3


Advanced Member
***

Group: Members
Posts: 205
Joined: 13-December 06
Member No.: 20,878
VR User?: 15 years
Which Program?: Dragon NaturallySpeaking 9.5 Preffered, Legal, Medical
Why do you Use VR?: SR consultant
Where are you Located: Howell New Jersey - USA
Other Comments: Former Senior Technical Solutions & Program Manager for Dragon NaturallySpeaking and the DNS Software Development Kit (SDK)




QUOTE(arcuhtek @ Sep 9 2008, 03:53 PM) [snapback]219492[/snapback]
Hello all!

I am new to VR software (and digital recording) so please bear with me. My question is simple, so I will ask first then give you some background.

I want to use a digital recorder (Olympus Ws-311M) to record meeting minutes in a conference room with between 4 and 10 people, then be able to transfer the resulting file to my PC and use a VR software to have the meeting converted into a Microsoft Word file that I can polish up and distribute to attendees. Is this possible with any version of DNS or other VR software?

Background.

Before you say search, I have already done so and read a few threads, which seemed to direct each poster to other threads about learning how to speak correctly and clearly or basically train their speech patterns such that the spoken statements interfaced most optimally with the software (DNS).

That is not possible for me, as per what I am suggesting, because anyone in the meeting will be speaking (I am sure some speech will be jumbled as people talk over one another), and I cannot reasonably expect everyone to speak with a "trained voice" for my own purposes.

I already have the recorder purchased and have recorded one 2 hour meeting and successfully transfered it to my PC. It is awaiting my next step of purchasing software to do the automatic translation OR......I can do it the old fashion way - put my ear phones on and manually translate it which will take me DAYS to do at my slow 40-50 wpm typing rate.

My expectation of a VR software is that it understand and translate 75% of the spoken words correctly and I do not even care about formatting. I fully expect to have to correct some words, insert a persons name beside each of their spoken statements and I am really fast at MS Word in terms of software efficiency, formatting, distribution, etc. (just not a fast "typist").

I would love to be able to use this method of taking meeting minutes often.

Please know I am only looking to anyone on this forum to give me a general idea if DNS (or other) software is purchased, that it can very basically covert meeting minutes into some form of digital text. I am not looking for a miracle. Does anyone recommend a free trial of a software that might accomplish my tasks? I hate to spend a lot of money just to try it out on one file.

Thanks!!


First, what you want to do and the way that you want to do it is stretching the capability of transcription to the absolute limits. It is highly unlikely that you will be able to directly transcribe with any degree of accuracy at all by allowing speech recognition software to transcribe speech to text from the audio file. There are some things that you can get away with. For example, with Dragon NaturallySpeaking versions 9.5 or 10.0, but best with 10.0, you can sometimes transcribe a one-on-one interview between yourself and another individual by creating a user profile that uses the speaker independent Acoustic Model provided with Dragon NaturallySpeaking. However, this takes some skill and is limited in terms of its overall effectiveness.

Second, in a conference room with 4 to 10 people all generally speaking conversationally and tending to talk over top of one another, this approach simply will not work. I could give you all of the detailed explanations on techniques and how to use them, but you would be spinning your wheels because: (1) the speakers would be way too far away from the microphone in the digital voice recorder to provide the clarity of recording required of speech recognition, (2) you can't use any type of "conference mode"on your digital voice recorder because such would not only amplify the speakers, but would amplify all the background noise etc., which would result in virtually no accuracy (garbage), (3) speech recognition requires structured speech (i.e. writing style/dictation) and does not work at all with conversational speech. Note that most conferences or meetings with numerous people are predominantly conversational vs. structured types of speech. These three factors by themselves would produce results that would be incoherent, not to mention virtually uneditable.

However, don't fret quite yet. These are the types of circumstances, under which most of those who post on these forums who understand how to do dictation and transcription using digital voice recorders, recommend is what is called "Echo dictation". This works quite well for users who do not type well. The way that this method works is as follows:

1. You listen to the recording of the conference using an earphone so that the audio from the conference doesn't play out loud where the speech recognition software can hear it. If you don't do this, and regardless of whether you're using a noise canceling microphone, speech recognition software may try to transcribe what it hears into text on your screen in your document. Therefore, you want to be able to listen to the recording without interfering with the SR software.

2. I'm going to address Dragon NaturallySpeaking because this is the most effective SR software application engaging in this type of transcription because you can format your text as you go along. While you're typing is slow, listening to the recording of your conference and then dictating into Dragon NaturallySpeaking as you go is up to three and four times faster than you can type, it doesn't misspell words, it does make some mistakes, but you will generally find that you can cut down the amount of time required for transcribing meetings and conferences by a significant and time-saving amount. There are many users out there who do this regularly. Dragon NaturallySpeaking allows you to dictate, with experience, up to 140 words per minute (practically) and up to 160 words per minute for those who are well trained in clear speech and enunciating. Dragon NaturallySpeaking also does not make spelling errors and frees you up to listen to what's being said. The entire process done in this way is far faster than trying to listen to a recorded conference or meeting and then trying to type. Typing under this condition splits your focus. However, it is very easy to listen to a sentence or two, dictate it, and then continue. I have demonstrated to clients how well this works, and I also am able to transcribe an hour's worth of audio in approximately an hour and 30 minutes (something that most any user can do after they have committed the process to instinct and do it automatically).

3. If you were to try to get Dragon NaturallySpeaking to transcribe the audio file directly, you would find that, next to it being impossible, even if you got 10 to 15% accuracy, it would take you twice as long as it does to transcribe and correct as it does currently for you to li listen and type.

Regardless of all of this, Echo dictation used with Dragon NaturallySpeaking is the best you will be able to do, but it will be significantly faster than what you currently do.

Chuck Runquist
Former Dragon NaturallySpeaking SDK & Senior Technical Solutions PM for DNS

"Life's Rule #1: Once you pull the pin, Mr. Grenade is no longer your friend." (Variant of Murphy's Law - Edward A. Murphy, Jr)
Go to the top of the page
 
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 30th October 2014 - 10:08 AM

We Recommend Using Dragon Medical and Dragon NaturallySpeaking Speech Recognition Software

Physicians Using Dragon Medical - Looking for the Best Electronic Health Record (EHR / EMR)?