KnowBrainer Speech Recognition
Decrease font size
Increase font size
Topic Title: Digital recorder settings
Topic Summary:
Created On: 03/04/2018 10:16 PM
Status: Post and Reply
Linear : Threading : Single : Branch
 Digital recorder settings   - cbrons - 03/04/2018 10:16 PM  
 Digital recorder settings   - cbrons - 03/04/2018 10:18 PM  
 Digital recorder settings   - Lunis Orcutt - 03/05/2018 10:41 AM  
 Digital recorder settings   - DaveR - 03/12/2018 01:36 AM  
 Digital recorder settings   - ErdaWilt - 12/29/2020 10:05 AM  
Keyword
 03/04/2018 10:16 PM
User is offline View Users Profile Print this message

Author Icon
cbrons
New Member

Posts: 9
Joined: 12/31/2012

I have a SONY ICD PX370 digital recorder. I am curious to know what settings you think would yield the highest accuracy when using Dragon Individual 15 to transcribe:

 

- For "Record Mode" it gives me the option of MP3 192kbps, MP3 128kbps, or MP3 48kbps (MONO). I have selected 48kbps. Is this best?

- For "Recording Filter" it gives me the option of NCF (Noise Cut), LCF (Low Cut), or OFF. I have selected OFF. Curious if this is the best option.

- For "VOR" it gives me the option of "ON" or "OFF". I don't even know what this is, so I have OFF selected.

- For "Built-in Mic Sensitivity" there are the options of HIGH, MEDIUM, LOW, and AUTO. I have LOW selected because I am just dictating my own voice notes.

 03/04/2018 10:18 PM
User is offline View Users Profile Print this message

Author Icon
cbrons
New Member

Posts: 9
Joined: 12/31/2012

Also, would it be better to buy a different digital recorder that produces files in another format? I.e. not MP3.
 03/05/2018 10:41 AM
User is offline View Users Profile Print this message

Author Icon
Lunis Orcutt
Top-Tier Member

Posts: 38677
Joined: 10/01/2006

The SONY ICD PX370 is not ideally suited or recommended for speech recognition. That's why you won't see it in the digital recorder section of our store.


MP3 is not ideally suited because it loses some of the recording. We only recommend lossless compression such as DS2 which is not available on any Sony recorder. We additionally dislike the recorder making a new recording every time you hit record and lack of glitchless punch in. If you already own this recorder, look for a WAV file (non-compressed) recording setting rather than MP3. The reason we prefer DS2 algorithms is because it is designed specifically for speech recognition (monophonic, limited frequency and lossless 12 to 1 compression ratio). While WAV files capture everything, Dragon has to filter out the higher frequencies. Sony recorders are designed for consumer music use. This doesn't mean that you recorder won't be adequate but it also means that it won't be ideal. Now to your questions…


- If you're going to use MP3, use the highest monophonic setting


- Recording filters special effect should be turned off (disabled)


- Avoid VOR which can result in cutting off the 1st syllable or word in each sentence


- Low (close mic mode) is your best selection


The Philips DPM-8000 is the current king of the hill but also somewhat expensive so it probably depends on if a more expensive recorder will justify the cost in your workplace. 



-------------------------

Forum Mission Statement
Trial Downloads
Dragon/Sales@KnowBrainer.com 
(615) 884-4558 ext 1

 03/12/2018 01:36 AM
User is offline View Users Profile Print this message


DaveR
New Member

Posts: 16
Joined: 10/08/2008

Notwithstanding the optimisation of DS2 for speech recognition, referred to by Lunis, my experience has been that there is no practical difference between MP3 and DS2 for recognition accuracy by Dragon. This is based on dictation into either an Olympus DS7000 (using DS2 - i.e. QP mode) or an Olympus DS3500 (which can record as MP3 128 or 256 kbps in addition to DS2), for recognition by Dragon DPI 15.

I've been doing this an hour a day for several years. Mostly I use the DS7000, on VCVA mode, but for meetings and interviews the DS3500 gives much better quality on MP3 (even 128 kbps), and sometimes I use that for dictation. The only drawback I find with MP3 is that you can't append the next dictation session - you have to start a new file, which makes organization harder. The fact that it's stereo doesn't seem to make any difference. These days if I'm not getting 100% accuracy, I can always trace it to one of the factors below.

The trick I found with VCVA (or VOR) mode is to get the sensitivity setting right, and there's a tension here with the gain of the microphone. Since I dictate in noisy environments (car, hospital ward), I like to use a low-gain microphone, but VCVA needs to be set to maximum sensitivity to avoid chopping off the initial syllable, as described by Lunis. On the Olympus recorders this is a sensitivity of 15. There's no fine-tuning of microphone sensitivity for these recorders, but adjusting the VCVA sensitivity, & enunciating clearly, eliminated the problem of chopping off the first syllable.

In fact there's not a huge advantage to using VCVA, because the recent versions of Dragon are very effective at filtering out background noise (occasionally too effective, as when a loud interfering noise renders Dragon operationally deaf for a period, which can be up to several minutes - I've sometimes lost whole tracts of dictation due to this). But what Dragon is not good at is filtering out noise from other people talking in the background, which is where VCVA can help.

In case it's thought this is a problem that could be solved by a more directional microphone, I should say I've spent many thousands on microphones of various kinds, before concluding the perfect microphone for this doesn't exist. The best by far for me is the Sennheiser ME3 headset, but a headset is clumsy in a hospital ward. The Samson Micro Airline is next best, but is fairly fragile, needs recharging, and can't be used on its lowest gain setting with VCVA. With the Samson you also have to be careful about the quality of the patch cable connecting the receiver to the recorder - anything but the best quality plugs tend to generate loud electrical interference when moved in the sockets, which can lead to the dictation loss described above.


-------------------------

DaveR

 12/29/2020 10:05 AM
User is offline View Users Profile Print this message

Author Icon
ErdaWilt
New Member

Posts: 1
Joined: 12/29/2020

Hi....I suggest setting it on mp3 as opposed to wav as it takes less space and is viewed as the standard configuration. I set it at 192kpbs,which isn't the most elevated setting.

Concerning situation with a fabulous piano, I place the recorder around 12 crawls over the piano 1 ft away close to the bend of the piano pointed toward the bass strings. Attempt this and see your opinion about the outcomes.

 

pcb assembly in usa



KnowBrainer Speech Recognition » Transcription and Digital Recorders » Digital recorder settings

Statistics
31988 users are registered to the KnowBrainer Speech Recognition forum.
There are currently 2 users logged in.
The most users ever online was 12124 on 09/09/2020 at 04:59 AM.
There are currently 515 guests browsing this forum, which makes a total of 517 users using this forum.

FuseTalk Standard Edition v4.0 - © 1999-2021 FuseTalk™ Inc. All rights reserved.