我不得不承认,当我被要求做教程时,我已经很长时间没有使用任何类型的语音识别了。(speech recognition)上次我尝试语音识别(speech recognition)是用Dragon Dictate,当 Windows 98 也是新的时候,它是新的。很久以前在计算机时代!
我很高兴解决这个问题,因为我很想看看语音识别是如何改进的。男孩,有没有改善!即使是像这样的基本内置应用程序,“开箱即用”也做得非常好。在这篇关于语音识别(Speech Recognition)的最后一篇文章中,我想谈谈我在重新认识语音识别(speech recognition)的奇迹时所学到的东西,以及我将从这里走向何方。
还没有准备好星际迷航
我敢肯定,我们中的许多人都看过星舰企业(Starship Enterprise)号的船员说"Computer!"并立即得到答复。我们还没有Starfleet计算机,但从Windows Vista开始,一直到Windows 7,我们确实有计算机会倾听我们的声音并响应我们告诉他们的内容——如果“那是什么?” ,就会回答我们。("What was that?")被认为是一个答案。
当我第一次尝试使用Speech Recognition时确实遇到了问题,并且故障排除并不是特别简单。帮助文件并不总是足够有用。我能够在Microsoft的网站(web site)和各种在线论坛上找到答案,而无需做太多工作。这就是我发现我忘记了我的网络摄像头(就在我面前的显示器顶部,直接与我说话的方式一致)也有一个有源麦克风的原因,这增加了混乱。一旦我解决了这个问题,从那时起它就非常顺利。
我什至尝试用各种不同的口音说话(例如BBC 英国和美国乡下人(BBC British and American redneck)),并且能够获得相当好的识别,允许标准美国发音的差异。当然是说"Friends, Romans, countrymen, lend me your ears!"虽然我对Helen Mirren 和 Jeff Foxworthy(Helen Mirren and Jeff Foxworthy)的印象最好,但让我笑得太多,无法得到完全准确的结果。
语言识别
语音识别(Speech Recognition)可以与不同的语言一起使用,我想我可以用我有限的美式(American)口音、西班牙语(Spanish)、德语和法语进行尝试,但不幸的是,除非您的(German and French)操作系统(operating system)也是该语言,否则您无法使用其他语言。您可以通过从Microsoft安装另一个语言包(language pack)来更改操作系统(operating system)的语言,但只有在运行Windows 7 Ultimate或Windows 7 Enterprise 时才能这样做。
语音识别(Speech Recognition)适用于美国英语(US English)、英国英语(UK English)、法语(French)、西班牙语(Spanish)、德语(German)、日语(Japanese)、繁体中文和简体中文(Traditional Chinese and Simplified Chinese),并将在这些语言的Windows 7版本(所有版本)中找到。我很抱歉不能尝试一下。我不知道Windows 8会发生什么,但我认为安装其他语言包的能力将是对Windows 7 Professional及更高版本的Windows 8等效项的一个很好的补充。
什么效果好
正如我所提到的,语音识别(Speech Recognition)旨在与其他Microsoft 软件(Microsoft software)配合使用。只要我尝试过Microsoft产品,我就非常成功(尽管正如人们所预料的那样,使用Microsoft Office Excel既有限又复杂)。使用其他软件时,它会被击中或错过。我可以很好地使用Google Chrome浏览器(肯定不如Internet Explorer)和我的Eudora电子邮件程序,它现在已经是非常古老的软件(antique software)了。值得尝试使用您自己喜欢的软件,看看您能做什么。“显示数字”命令在选择项目和命令时特别有用。
我还发现,识别的准确率并没有花很长时间就显着提高。我进行了两次训练练习,之后识别几乎100 % correct。我能够说得快一点,并减少停顿让软件跟上。我真的很喜欢看我的声音在屏幕上翻译成文字。我早期使用语音识别软件(speech recognition software)的体验远非如此愉快。
什么不好用
正如我所提到的,某些软件与Speech Recognition不兼容。我什至无法打开Adobe Reader或TweetDeck的(TweetDeck)Adobe AIR版本。我发现我无法使用Internet Explorer登录我的(Internet Explorer)Google 帐户(Google account)来试用Google(Google Docs) Docs——似乎无法说出或拼写我的密码。我怀疑这是一个安全问题(security issue),不允许在其他人可能听到的地方大声说出密码,但这很烦人。
我可以打开iTunes并选择要播放的歌曲,但实际上无法播放。我可以打开Scrivener(我选择的文字处理器),但“显示数字”("Show numbers")并没有在我想使用的任何东西上覆盖数字。我没有对我最喜欢的软件进行任何真正广泛的实验——这些只是我尝试过的几个。对于任何想要使用语音识别(Speech Recognition)来测试他们想要使用它的程序的人来说,这是值得的,以确保它是兼容的。
更多链接和资源
如果你还没有看过之前的文章,你可以在这里找到它们:
- 使用语音识别(Speech Recognition):设置和配置(Setup and Configuration)
- 使用语音识别(Speech Recognition):命令、听写文本和浏览(Dictating Text & Browsing)网页(Web)
- 解决常见的语音识别问题(Common Speech Recognition Problems)
奇怪的是,如果不单击Google 或 Bing 搜索(Google or Bing search)中的链接,几乎不可能在Microsoft Answers网站(web site)上找到任何有关语音识别(Speech Recognition)的信息。将“语音识别”("Speech Recognition")放入搜索框(search box),我根本无法得到任何答案,尽管论坛中有一些关于它的问题。使用此链接从 Windows 网站获取语音识别帮助:(Speech Recognition)语音识别搜索结果(Speech Recognition search results)。
这是一篇关于Microsoft语音(Microsoft)识别(speech recognition)历史的简短Wikipedia文章(Wikipedia article):Windows Speech Recognition。
这是一篇博客文章(blog entry),提供了作者关于将语音识别(Speech Recognition)与Dragon Naturally speak(Dragon Naturally Speaking)进行比较的想法:Dragon NaturallySpeaking 与 Windows 7 语音识别(Dragon NaturallySpeaking Versus Windows 7 Voice Recognition)。
把它包起来
我真的很喜欢使用语音识别(Speech Recognition),并惊叹于随着时间的推移所取得的改进。对于日常休闲使用(everyday use)来说,它肯定足够好,尤其是对于微软(Microsoft)产品。
我会继续使用语音识别(Speech Recognition)吗?是的,当我可以的时候。在这一点上,我不需要任何更复杂的东西。花时间训练它并训练自己正确使用它是非常值得的。
Speech Recognition: Afterthoughts on Its Strengths and Weaknesses
I have to аdmit that I hadn't used any kind of speech recognition in a long time when I was asked to do the tutorials. The last time I had tried speech reсоgnition, it wаs with Dragon Dictate, which was new when Windows 98 was also new. A very long time ago in computer years!
I was happy to tackle this subject because I was very interested to see how speech recognition has improved. And boy, has it improved! Even a basic, built-in application like this did an amazingly good job "right out of the box." In this final article about Speech Recognition, I'd like to talk about what I learned while reacquainting myself with the wonders of speech recognition, and where I'll be going from here.
Not ready for Star Trek yet
I'm sure many of us watched the crew of the Starship Enterprise saying "Computer!" and getting an immediate answer. We don't have Starfleet computers yet, but beginning with Windows Vista and continuing in Windows 7, we do have computers that will listen to us and respond to what we tell them—and answer us, if "What was that?" is considered an answer.
I did have problems when I first tried to use Speech Recognition, and troubleshooting wasn't particularly straightforward. The help files are not always helpful enough. I was able to find the answers on Microsoft's web site and in an assortment of online forums without too much work. That's how I found out that I had forgotten that my webcam (sitting right in front of me on top of my monitor, and directly in line with the way I was speaking) also had an active microphone and was adding to the confusion. Once I fixed that, it was pretty smooth sailing from then on.
I even tried speaking with an assortment of different accents (BBC British and American redneck, for example) and was able to get reasonably good recognition, allowing for differences in standard American pronunciation. Of course, saying "Friends, Romans, countrymen, lend me your ears!" while doing my very best impressions of Helen Mirren and Jeff Foxworthy made me laugh too much to get entirely accurate results.
Language recognition
Speech Recognitioncan be used with different languages, and I thought I might try it out with my limited, American accented, Spanish, German and French, but unfortunately you cannot use other languages unless your operating system is also in that language. You can change your operating system's language by installing another language pack from Microsoft, but you can only do that if you're running Windows 7 Ultimate or Windows 7 Enterprise.
Speech Recognition is available for US English, UK English, French, Spanish, German, Japanese, Traditional Chinese and Simplified Chinese, and will be found in those languages' versions of Windows 7 (all versions). I was sorry not to be able to try that out. I have no idea what will happen with Windows 8, but I think the ability to install other language packs would be a good addition to the Windows 8 equivalents of Windows 7 Professional and above.
What works well
As I mentioned, Speech Recognition is designed to work best with other Microsoft software. As long as I experimented with Microsoft products I was very successful (although as one might expect, using Microsoft Office Excel was both limited and complicated). With other software it was hit or miss. I could use the Google Chrome browser fairly well (definitely not as well as Internet Explorer) and my Eudora email program, which is pretty much antique software by now. It's worth experimenting with your own favorite software to see what you can do. The "show numbers" command was especially helpful in selecting items and commands.
I also found that it didn't take very long for the accuracy of the recognition to improve markedly. I went through the training exercises twice, and after that the recognition was almost 100% correct. I was able to speak a little faster and put in fewer pauses for the software to keep up. I really enjoyed watching my voice translated into words on the screen. My early experiences with speech recognition software were nowhere near this pleasant.
What doesn't work well
As I mentioned, some software is just incompatible with Speech Recognition. I couldn't even open Adobe Reader or the Adobe AIR version of TweetDeck. I found that I could not sign into my Google account with Internet Explorer to try out Google Docs—there seemed to be no way to speak or spell my password. I suspect this is a security issue, not allowing passwords to be spoken out loud where someone else might hear, but it was annoying.
I could open iTunes and select a song to play, but could not actually get it to play. I could open Scrivener (my word processor of choice) but "Show numbers" did not overlay numbers on anything I wanted to use. I didn't do any really extensive experimentation with my favorite software—those are just a few that I tried. It would be worthwhile for anyone who wants to use Speech Recognition to test out the programs they want to use it with, to be sure it's going to be compatible.
More links and resources
If you haven't already seen the previous articles, you can find them here:
Oddly, it's almost impossible to find any information about Speech Recognition on the Microsoft Answers web site without clicking a link from a Google or Bing search. I was unable to get any answers at all by putting "Speech Recognition" into the search box, even though there are a few questions about it in the forums. Use this link to get Speech Recognition help from the Windows web site: Speech Recognition search results.
Here's a brief Wikipedia article that talks about the history of speech recognition at Microsoft: Windows Speech Recognition.
Here is a blog entry that gives the author's thoughts on comparing Speech Recognition with Dragon Naturally Speaking: Dragon NaturallySpeaking Versus Windows 7 Voice Recognition.
Wrapping it up
I really enjoyed working with Speech Recognition and marveling at the improvements that have been made over time. It would certainly be good enough for casual everyday use, especially with Microsoft products.
Will I keep using Speech Recognition? Yes, when I can. At this point I don't need anything more sophisticated. It was well worth the time it took to train it and to train myself to use it right.