English Speech Files

Flat
ralfherzog-20070813_vf12.zip
User: ralfherzog
Date: 8/12/2007 9:02 pm
Views: 6758
Rating: 15
Hello everyone! This is my fourth contribution to the VoxForge project. Wink I used my nearly-new <a href="http://www.sennheisercommunications.com/comm/icm_eng.nsf/root/05349" title="Sennheiser PC 131">Sennheiser PC 131</a> headset which has a noise canceling microphone. It should have less spikes than my previous contribution which I made with a cheap Trust microphone.  But I am not happy with the quality of my recordings - I want to improve, I am thinking about buying a new PCI sound card instead of using the onboard sound card. If you have any suggestions about a not too expensive, but good PCI soundcard, please let me know.

Speaker Characteristics:

Gender: male;
Age range: adult;
Pronunciation dialect: General American.

Recording Information:

Microphone make: Sennheiser PC131;
Microphone type: Noise canceling headset;
Audio card make: OnBoard - Mainboard: Asrock 939Dual-SATA2;
Audio card type: OnBoard - Realtek 850;
Audio Recording Software: Audacity 1.2.6;
O/S: Windows XP Professional.

File Info:

File type: wav;
Sampling rate: 48kHz;
Sample rate format: 16bit;
Number of channels: [1];
Audio Processing: [y/n] n

vf12-01 Violent life and athletic sports had never appealed to me
vf12-02 You live on an income which your father earned
vf12-03 He was worth nothing to the world
vf12-04 Then you don't believe in altruism
vf12-05 The creative joy, I murmured
vf12-06 He deluged me, overwhelmed me with argument
vf12-07 Ah, it is growing dark and darker
vf12-08 I was Hump, cabin boy on the schooner Ghost
vf12-09 A sinewy hand, dripping with water, was clutching the rail
vf12-10 No man ate of the seal meat or the oil
vf12-11 I noticed blood spouting from Kerfoot's left hand
vf12-12 Three oilers and a fourth engineer, was his greeting
vf12-13 Eighteen hundred, he calculated
vf12-14 The sharp voice of Wolf Larsen aroused me
vf12-15 I obeyed, and a minute or two later they stood before him
vf12-16 But it won't continue, she said with easy confidence
vf12-17 What I saw I could not at first believe
vf12-18 The stout wood was crushed like an eggshell
vf12-19 There's too much of the schoolboy in me
vf12-20 I had forgotten their existence
vf12-21 Ah, we were very close together in that moment
vf12-22 But she swung obediently on her heel into the wind
vf12-23 They are his tongue, by which he makes his knowledge articulate
vf12-24 Between the rush of the cascades, streaks of rust showed everywhere
vf12-25 He'll never do a tap of work the whole Voyage
vf12-26 Captain West may be a Samurai, but he is also human
vf12-27 And so early in the voyage, too
vf12-28 In the matter of curry she is a sheer genius
vf12-29 The eastern heavens were equally spectacular
vf12-30 He spat it out like so much venom
vf12-31 I saw Mr Pike nod his head grimly and sarcastically
vf12-32 He is too keenly intelligent, too sharply sensitive, successfully to endure
vf12-33 The night was calm and snowy
vf12-34 I sailed third mate in the little Vampire before you were born
vf12-35 His outstretched arm dropped to his side, and he paused
vf12-36 At this moment I felt a stir at my shoulder
vf12-37 Wada, Louis, and the steward are servants of Asiatic breed
vf12-38 Also, she has forbidden them smoking their pipes in the after-room
vf12-39 I tried to read George Moore last night, and was dreadfully bored
vf12-40 Tom Spink has a harpoon  

Copyright (C) 2007  Ralf Herzog

These files are free software; you can redistribute them and/or
modify them under the terms of the GNU General Public License
as published by the Free Software Foundation; either version 2
of the License, or (at your option) any later version.

These files are distributed in the hope that they will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

--- (Edited on 8/12/2007 9:02 pm [GMT-0500] by ralfherzog) ---

ralfherzog-20070813_vf12.zip ralfherzog-20070813_vf12.zip

Notice: many prompts in "English Speech Files" were adapted from the prompt files contained in the CMU_ARCTIC speech synthesis database, which were in turn derived from out-of-copyright texts from Project Gutenberg, by the FestVox project at the Language Technologies Institute at Carnegie Mellon University.

Re: ralfherzog-20070813_vf12.zip
User: Robin
Date: 8/13/2007 6:14 am
Views: 152
Rating: 15

Perhaps in the forums more people will notice this question. No soundcard suggestions, but one other option to consider perhaps: a USB soundcard.

An advantage: same sound system on every computer you work on even on laptops (good if you work with speaker dependant speech recognition, because your soundcard can alter your voice so your acoustic model needs to be adapted to the combination mic+card).

Disadvantages: might be hard to get to work with some programmes (though this will only come into play under linux I guess), might draw a tiny bit more power on a laptop?

--- (Edited on 8/13/2007 6:14 am [GMT-0500] by Robin) ---


Notice: many prompts in "English Speech Files" were adapted from the prompt files contained in the CMU_ARCTIC speech synthesis database, which were in turn derived from out-of-copyright texts from Project Gutenberg, by the FestVox project at the Language Technologies Institute at Carnegie Mellon University.

Re: PCI Sound Card Recommendations
User: kmaclean
Date: 8/13/2007 9:29 am
Views: 140
Rating: 14

Cross posted to Audio Forum

Ken 

--- (Edited on 8/13/2007 10:29 am [GMT-0400] by kmaclean) ---

--- (Edited on 8/13/2007 10:30 am [GMT-0400] by kmaclean) ---


Notice: many prompts in "English Speech Files" were adapted from the prompt files contained in the CMU_ARCTIC speech synthesis database, which were in turn derived from out-of-copyright texts from Project Gutenberg, by the FestVox project at the Language Technologies Institute at Carnegie Mellon University.

Re: ralfherzog-20070813_vf12.zip
User: kmaclean
Date: 8/13/2007 9:19 am
Views: 176
Rating: 19

Hi Ralf,

Your last two submissions are now incorporated into the VoxForge Corpus and Acoustic Model.

Thanks!

Ken 

--- (Edited on 8/13/2007 10:19 am [GMT-0400] by kmaclean) ---


Notice: many prompts in "English Speech Files" were adapted from the prompt files contained in the CMU_ARCTIC speech synthesis database, which were in turn derived from out-of-copyright texts from Project Gutenberg, by the FestVox project at the Language Technologies Institute at Carnegie Mellon University.

Re: ralfherzog-20070813_vf12.zip
User: Robin
Date: 9/21/2007 7:48 am
Views: 174
Rating: 14

Hi Ralph,

I have a question in that was rather hard to answer using information on websites, but perhaps you can. I have a Sennheiser headset (pc146) and starting Monday I will work with Dragon Naturally Speaking. I think there is also a headset included with the program, so I was wondering whether I should use my own headset or that one. some people say that the included headset isn't very good, but obviously Nuance says that it is good. I just wonder whether mine is just as good or better in which case I will probably use mine.

thanks,

Robin 

--- (Edited on 9/21/2007 7:48 am [GMT-0500] by Robin) ---


Notice: many prompts in "English Speech Files" were adapted from the prompt files contained in the CMU_ARCTIC speech synthesis database, which were in turn derived from out-of-copyright texts from Project Gutenberg, by the FestVox project at the Language Technologies Institute at Carnegie Mellon University.

Dragon headset, Sennheiser PC 146, Andrea USB soundcard
User: ralfherzog
Date: 9/21/2007 12:24 pm
Views: 2458
Rating: 19
Hello Robin,

Short version: You should continue to use your Sennheiser headset PC 146.

Long version: When you start using NaturallySpeaking, NaturallySpeaking will make an adjustment of the loudness of your microphone.  So NaturallySpeaking can adjust the loudness of your Sennheiser PC 146 headset.

After NaturallySpeaking has adjusted the loudness of your headset, you can start using Audacity 1.2.6.  Audacity will use the same loudness as previously adjusted by NaturallySpeaking.  So the setting for the loudness of the microphone should be optimal.

Then you can record your own voice with Audacity (after making the adjustment of the microphone level by NaturallySpeaking).  Record your voice with your Sennheiser PC 146 headset.

Afterwards, make the same procedure using the Dragon headset.  Adjust the volume setting with NaturallySpeaking using your Dragon headset.  And then make a record with Audacity.

Compare both recordings (the first one with your Sennheiser PC 146 headset and the second one with your Dragon headset) you made with Audacity.  Probably, the recording with the Sennheiser PC 146 headset is a bit better.

What is the reason for the assumed difference of quality?  In my opinion, the main reason is probably the quality of the sound card.  Your Sennheiser PC 146 headset has an integrated USB soundcard, which is probably better than your onboard sound card.

If you want to use your Dragon headset, you should consider to buy the Andrea USB adapter (here one offer at eBay's auction house: http://tinyurl.com/36vg7v - the price is about 25 €).  I am using this Andrea USB adapter.  And it is worth the price for speech recognition enthusiasts like we are.  

The Andrea USB adapter works under my Ubuntu Linux operating system.  They don't advertise that the Andrea USB adapter is working under Linux, but in my case, it is working.  So this might be an additional reason to buy the Andrea USB adapter.  It has a good compatibility and a good quality for speech recordings.

But once you have started using NaturallySpeaking, you probably won't use Linux anymore because you don't want to miss NaturallySpeaking anymore.

An alternative to the suggested Andrea USB soundcard might be the following soundcard "Buddy USB 5G™ Sound Adapter" - take a look here: http://tinyurl.com/3yveve

Before you buy the Andrea USB adapter, you might want to have a look at this discussion: http://tinyurl.com/2nfcy9

If you don't want to buy the Andrea USB adapter, you should take a look at what soundcards Dragon NaturallySpeaking is officially recommending.  They do recommend some specific soundcards.  This information should be written on the package of NaturallySpeaking.  You can buy one of those soundcards they are referring to.  But don't buy any other soundcard, the result might be disappointing.

The third alternative would be to buy a completely new mainboard with a good onboard sound card.  Probably, the quality would be sufficient.  Because new mainboards tend to have a much better sound quality than older main boards.

In my opinion, a soundcard for speech recognition shouldn't cost more than 35 €.  There are professional recording cards available (example: M-Audio Audiophile 2469, price about 80 €).  But I'm not sure whether such an expensive sound card is necessary.  I wouldn't spend more than 35 € for a soundcard that should be used for speech recognition.  I am sure that it would be a waste of money to buy a recording card for 80 €.  But I haven't tested it.  But I have read a lot, and I think that the maximum price for a recording soundcard for speech recognition is 35 €.

You see, this is a complex topic. I have read a lot about it in the Internet to find the best solution.  In my opinion, you should focus on the soundcard, not on the microphone.  And a good choice is the Andrea USB adapter, I am using this Andrea USB adapter, you can listen to my recordings to get an impression of the quality.  There is little background noise, you can live with that background noise.

The Andrea USB adapter has the advantage that you can buy different normal headsets with no integrated USB soundcard.

You can have the best microphone, but without the appropriate soundcard you won't get a good result.

You can forget USB soundcards that cost about 10 €.  They are too bad.

So an optimal price/quality solution should be a headset for about 35 € and a soundcard (PCI or USB) for not more than 35 €.  It shouldn't be more expensive.  And it shouldn't be less expensive.  If you would pay more, you would waste your money.  If you would pay less, your results might be too bad.

One additional question: Which version of NaturallySpeaking are you using?  The standard version or the preferred version?  And which languages are you able to use?

And don't forget to update NaturallySpeaking!  NaturallySpeaking 9.1 (Windows XP) is much better than NaturallySpeaking 9.  If you're using Windows Vista (I don't use Vista), there should be NaturallySpeaking 9.5 available, which is probably a bit better than NaturallySpeaking 9.1.  It is necessary that you update to the newest version via Internet to get the best recognition results.

If you have questions, concerning soundcard, microphone or NaturallySpeaking, please ask.  I am very interested in this topic and I would like to find a better solution for myself than I am having today.

You can test the Dragon headset in the combination with the Andrea USB adapter.  And compare the result with your Sennheiser PC 146 headset.  I am not sure which solution would be better.

You see, it is really difficult to find a good answer.

Greetings, Ralf

--- (Edited on 9/21/2007 12:24 pm [GMT-0500] by ralfherzog) ---


Notice: many prompts in "English Speech Files" were adapted from the prompt files contained in the CMU_ARCTIC speech synthesis database, which were in turn derived from out-of-copyright texts from Project Gutenberg, by the FestVox project at the Language Technologies Institute at Carnegie Mellon University.

Re: Dragon headset, Sennheiser PC 146, Andrea USB soundcard
User: Robin
Date: 9/28/2007 1:17 pm
Views: 500
Rating: 19

Hi Ralf,

I still needed to thank you for that information.  It was a busy work though, but your information came in very handy, because I needed to decide which which microphone to work.  I am now working with my own as you suggested and the results so far are quite impressive.

When I have more free time I will definitely experiment a bit more, but at the moment unfortunately time doesn't allow.


Thanks!
Robin

--- (Edited on 9/28/2007 1:17 pm [GMT-0500] by Robin) ---


Notice: many prompts in "English Speech Files" were adapted from the prompt files contained in the CMU_ARCTIC speech synthesis database, which were in turn derived from out-of-copyright texts from Project Gutenberg, by the FestVox project at the Language Technologies Institute at Carnegie Mellon University.

PreviousNext