<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0">
<channel>
<title>Recording Equipment Discussions</title>
<link>http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions</link>
<description></description>

<item>
<title>Samsung WEP 475</title>
<link>http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/samsung-wep-475</link>
<description>I picked up one of these for $20 at a local bargain electronics store. I was interested to see how well the noise cancellation worked. I have a fairly quiet location but it is plagued by noisy WiFi signals which cause a lot of popping, scratching and clicking on my regular headsets. </description>
<guid isPermaLink="true">http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/samsung-wep-475</guid>
<pubDate>Fri, 22 Jul 2011 10:51:34 -0500</pubDate>
</item>

<item>
<title>failed to begin input stream</title>
<link>http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/failed-to-begin-input-stream</link>
<description>All, </description>
<guid isPermaLink="true">http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/failed-to-begin-input-stream</guid>
<pubDate>Mon, 06 Jun 2011 05:22:14 -0500</pubDate>
</item>

<item>
<title>Small linux device for voice?</title>
<link>http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/small-linux-device-for-voice</link>
<description>One of the issues that concerns me when dealing with voice applications is that mainstream devices are overburdened with resources to deal with visual output. </description>
<guid isPermaLink="true">http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/small-linux-device-for-voice</guid>
<pubDate>Thu, 17 Mar 2011 06:10:26 -0500</pubDate>
</item>

<item>
<title>Speech Submission Applet Source</title>
<link>http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/speech-submission-applet-source</link>
<description>Hello folks, </description>
<guid isPermaLink="true">http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/speech-submission-applet-source</guid>
<pubDate>Fri, 03 Dec 2010 09:03:36 -0600</pubDate>
</item>

<item>
<title>Problem on Read for portuguese on voxforge</title>
<link>http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/problem-on-read-for-portuguese-on-voxforge</link>
<description>Hello folks, as I have not found any contact for the VoxForge team, I am reporting an error while trying to read for Portuguese, the Java applet in the main read page show something like this: </description>
<guid isPermaLink="true">http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/problem-on-read-for-portuguese-on-voxforge</guid>
<pubDate>Fri, 13 Aug 2010 07:53:11 -0500</pubDate>
</item>

<item>
<title>Number of sentences per form</title>
<link>http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/number-of-sentences-per-form</link>
<description>Hi everybody. </description>
<guid isPermaLink="true">http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/number-of-sentences-per-form</guid>
<pubDate>Mon, 19 Jul 2010 12:47:52 -0500</pubDate>
</item>

<item>
<title>Skip sentences</title>
<link>http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/skip-sentences</link>
<description>Hi everybody. </description>
<guid isPermaLink="true">http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/skip-sentences</guid>
<pubDate>Sun, 18 Jul 2010 07:37:10 -0500</pubDate>
</item>

<item>
<title>Recommendations on a good desktop microphone?</title>
<link>http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/recommendations-on-a-good-desktop-microphone</link>
<description>I have been finding that a speaker dependent model is also dependent on the microphone being used.  Does anybody have a recommendation regarding a good desktop microphone for speech recognition.  Hopefully one that would work well even without me leaning towards it to speak. </description>
<guid isPermaLink="true">http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/recommendations-on-a-good-desktop-microphone</guid>
<pubDate>Mon, 26 Apr 2010 11:56:05 -0500</pubDate>
</item>

<item>
<title>Jabra bt2040 and batteries</title>
<link>http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/jabra-bt2040-and-batteries</link>
<description>Some of my testing errors with this headset (which has a replaceable battery) appear to be due to the state of charge of the battery. The supplied AAAA battery is marked Alkaline 1.5V. I have been using rechargeable NiMh batteries of the same approximate size. When fully charged these appear to have about 1.35V. This is enough to run the headset but errors increase markedly when down to about 1.30V. </description>
<guid isPermaLink="true">http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/jabra-bt2040-and-batteries</guid>
<pubDate>Tue, 26 Jan 2010 06:15:57 -0600</pubDate>
</item>

<item>
<title>Single monolithic model, or separate models?</title>
<link>http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/single-monolithic-model-or-separate-models</link>
<description>One of the issues that arises when multiple audio input/output options are available is whether, theoretically, it is better to put all the eggs in one basket or to keep them separate. </description>
<guid isPermaLink="true">http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/single-monolithic-model-or-separate-models</guid>
<pubDate>Mon, 18 Jan 2010 10:47:28 -0600</pubDate>
</item>

<item>
<title>Sennheiser PC131 Wired</title>
<link>http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/sennheiser-pc131-wired</link>
<description>I picked up one of these for a couple of reasons: 1. Noise cancelling - sometimes extraneous sounds interfere with my recordings 2. Quality of incoming sound - sometimes it is hard to hear what Festival voices are saying and better quality headphones might improve that situation 3. Wired, so instant-on - wireless headsets take a small time to wake up </description>
<guid isPermaLink="true">http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/sennheiser-pc131-wired</guid>
<pubDate>Sat, 16 Jan 2010 10:04:59 -0600</pubDate>
</item>

<item>
<title>Using pocketsphinx_tidigits</title>
<link>http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/using-pocketsphinx_tidigits</link>
<description>   hello everyone </description>
<guid isPermaLink="true">http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/using-pocketsphinx_tidigits</guid>
<pubDate>Thu, 14 Jan 2010 23:39:04 -0600</pubDate>
</item>

<item>
<title>Wireless headsets</title>
<link>http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/wireless-headsets</link>
<description>Bluetooth: Jabra BT2040, Jabra VBT185Z, Plantronics Explorer 360 with Belkin USB dongle Logitech Clearchat Pro Wireless USB (comes with headset and dongle.) My purpose was to get a wireless set which would allow me to record prompts, interact with my dialog manager, and experiment several hours a day creating models while being able to walk around the room and do other things. The result overall is that all of them work equally well allowing me to achieve close to 100% recognition accuracy after a non-trivial setup process on my Linux box. But they have subtle differences. The Logitech is a comfortable wear, with padded earpieces, nice quality sound coming in, easy control of mike on and off with a big button, no volume control on the headset, most difficult to wear if you are in a cool room and wear a hoodie at the same time. This unit is relatively expensive, MSRP in Canada is about $130, I got mine on sale at about $80. Unless you are looking for the nicer sound and comfy padded headsets it does not give any advantage over bluetooth except for: - I did not notice any crackling on this unit, but I did on all the BT sets from time to time. - It does have a hum/hiss of its own quite different from the BT sets, which might give some advantage It does not have a noise cancelling mike, which normally would come in that price range. The bluetooth sets: They are all equally uncomfortable to wear for an extended period stuffed into your ear. However they work surprisingly well. The J2040 was $13, the J185 $15 and the Explorer $30. The Plantronics seems to be the most robustly built, the Plantronics and the J185 have tiny buttons for volume control which I could not use once it was in my ear. Sound quality is quite good, but used with Festival TTS there were times when I could not make out clearly individual words which might have been clearer through a quality speaker. My favourite is the Jabra 2040. It has no volume controls on the unit but that is irrelevant to me. It is the cheapest and the quality is fair. The big advantage is that it has a replaceable battery (AAAA type). This is really good when you are working away and the batt voltage disappears suddenly. Swap out the battery for a new one and you are on your way again with minor inconvenience. The others (including the Logitech) take about 2 hours to recharge which is a pain if you are on a roll. AAAA (four As) batteries can be hard to find. There is some advice on the Internet regarding using the component batteries of some rechargeable 9V batteries. I dismantled one of these and now have a stack of rechargeable batteries which can be swapped out in less than a minute. This process is not recommended for situations where the headset can be knocked about since the batteries are slightly smaller than the battery compartment which gives it room to move. When I get errors with wireless sets they tend to be the words ZERO, ONE, LIMA and SYSTEM. These are the words that pop up in my dialog manager diagnostics most frequently. Sometimes when I pronounce these words Julius thinks I am saying something else, but what Julius hears is different each time. This seems to say that the word enunciated is interfered with by cracks, pops, ticks, missing packets but quite randomly, otherwise Julius would hear the same thing each time. This makes it hard to have the dialog manager flag a possible error when a particular word is heard. </description>
<guid isPermaLink="true">http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/wireless-headsets</guid>
<pubDate>Fri, 08 Jan 2010 05:40:24 -0600</pubDate>
</item>

<item>
<title>A note on Linux hardware recognition</title>
<link>http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/a-note-on-linux-hardware-recognition</link>
<description>For all the hardware combinations I have tried so far, Windows has detected the device as an audio device and so there has been no problem selecting that device in the OS. I have not tested them all with Julius on Windows, however. </description>
<guid isPermaLink="true">http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/a-note-on-linux-hardware-recognition</guid>
<pubDate>Sun, 27 Dec 2009 08:39:40 -0600</pubDate>
</item>

<item>
<title>The importance of hardware selection</title>
<link>http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/the-importance-of-hardware-selection</link>
<description>This could be a very informative section. I look forward to contributing and learning from the experiences of others. Particulary for beginners with speech experimentation, if you can start out successfully with the right equipment then it is more likely your experience will be long and positive. More advanced users may come at the issue from the point of view that &#x22;I want to pick up any kind of hardware and use it despite its shortcomings, that is what is out there in the real world.&#x22; On the other hand, if you use your hardware a lot then you want something that is durable and reliable. It is not a pleasant experience to put time into creating a model and then have the hardware fail when you need to use it. Once I find a good combination I will probably buy up a crateload of them to ensure I have plenty of backups. There are other fora where the relative merits of hardware are discussed, but they rarely look at the hardware in sufficient detail relevant to speech experimentation. It is also very important to discuss the hardware in the context of the entire train, not just one component, as Robin has suggested above. Personally I have used 8 different devices for speech input and have found them all very odd in one way or another. Let me dismiss the 3 worst here: all my comments are based on an IBM NetVista using motherboard sound and USB 1 with Linux operating system, ALSA and Pulseaudio. 1. Karaoke type generic microphone plugs into the motherboard. Horrible DC offset in Audacity, it is usable but hard to listen to the result. Unplugged it and never used it again. 2. A desktop boom mike that shipped with a Creative AWE many years ago does not seem to work any more. Unreliable. 3. Labtec C-322 is a headset with a noise cancelling mike. This worked well for a while but could not stand up to the usage. The headband is not large enough for my skull and so the earpieces are only close to my ears, not over them. The foam earpadding for the earpieces wore off quite quickly - guys with prickly sideburns should watch for wear here - and the earpieces stopped working when the wiring failed at the plug end despite very careful handling. I will be using an adapter between the plugs and the actual headset with this type of setup in the future. That way all the mechanical load is on the adapter and not the plug, and the adapter is inexpensively replaceable. I put a lot of work into experiments with this headset, most of this is history now since I am not using that headset any more. Longevity is important. It is a pity about this headset, I managed quite a few successful 100% recognition models and still haul it out on occasion when a wired solution is good and I can spend time fixing it up when the earphones go sour. </description>
<guid isPermaLink="true">http://www.voxforge.org/home/forums/message-boards/recording-equipment-discussions/the-importance-of-hardware-selection</guid>
<pubDate>Sun, 27 Dec 2009 06:32:33 -0600</pubDate>
</item>

<item>
<title>New Microphones Sub-forum added</title>
<link>http://www.voxforge.org/home/forums/message-boards/microphone-discussions/new-microphones-sub-forum-added</link>
<description></description>
<guid isPermaLink="true">http://www.voxforge.org/home/forums/message-boards/microphone-discussions/new-microphones-sub-forum-added</guid>
<pubDate>Mon, 21 Dec 2009 13:32:36 -0600</pubDate>
</item>

</channel>
</rss>

