Speech technology

[jamesdaff-p40]

I'm curious if anyone knows of any interesting applications of speech technology within the church. I am aware of a few people who have talked about them, but not very many that have actually been used. Here are a few I know of, either actual or speculative:
1. I have heard of people using commercial speech recognition software to prepare talks.
2. I know of one ward where an enterprising member got the bishopric's permission to record the sacrament meeting talks digitally, for eventual audio indexing. (E.g. What was that talk last year where they talked about Elder Holland's missionary experience?)
3. I know of several members who use speech recognition to produce reasonable transcriptions within a church setting.

I'd like to know more about anything like this that other are aware of.

[idback9-p40]

Dragon Naturally Speaking has been in the business for some time now. A version of it is built into Microsoft Office. The new windows Vista has it built into the OS and is very fun to use in navigating and running commands. As far as text, recording church talks would not work because the program requires that you spend some time training it to your specific speech patterns. Using it to write with is also problematic in that it is only about 95% acurate. Fixing the other 5% seems to take about as much time as the speech seems to save. However, it is worth spending some time with. On a Windows XP box, right click the menu bar and add the Language Bar toolbar. Have fun and good luck.

[jamesdaff-p40]

I am a director of research for Dragon, so I know something about how it works and what can be done with it. Let me correct a couple of points:
1. The speech recognition in MS Vista is NOT Dragon, it is Microsoft's home-grown speech recognizer. Use it at your peril.
2. The most recent release of Dragon Naturally Speaking does not require training.
3. For some years now, Dragon has sold a speaker-independent version of the software for use in "audiomining" applications, such as searching church talks. The accuracy is less than what you'd get for a motivated dictator, but if you're looking for a talk on "tithing", you only have to recognize one instance of it in the talk to get a hit. Audio searching is a fairly well-established technology.

[james_francisco]

I am a director of research for Dragon, so I know something about how it works and what can be done with it. Let me correct a couple of points:
1. The speech recognition in MS Vista is NOT Dragon, it is Microsoft's home-grown speech recognizer. Use it at your peril.
2. The most recent release of Dragon Naturally Speaking does not require training.
3. For some years now, Dragon has sold a speaker-independent version of the software for use in "audiomining" applications, such as searching church talks. The accuracy is less than what you'd get for a motivated dictator, but if you're looking for a talk on "tithing", you only have to recognize one instance of it in the talk to get a hit. Audio searching is a fairly well-established technology.

Jamesdaff said:
1. The speech recognition in MS Vista is NOT Dragon, it is Microsoft's home-grown speech recognizer. Use it at your peril.

I don't think that MS may necessarily recommend it either. When I worked at MS Human resources IT a few years back we did speech recognition accessibility testing with Dragon, not the built in Speech API.

James Francisco

[reachtheworld-p40]

Many patriarchs use speech recognition programs to transcribe patriarchal blessings.

And speech tech the other way around -- having the computer read text out loud, has obvious potential Church-related uses. Don't know about Vista, but the upcoming Mac OS X is supposed to have a nice built-in text-to-speech engine.

[Roger-p40]

I tried using Dragon to transcribe patriarchal blessings but had touble. I had a older version but could not afford to upgrade so I gave it up. But it would save a lot of transcribing time if it would work.

Roger

[thedqs]

The upcoming Mac OS X is supposed to have a nice built-in text-to-speech engine.

Were you refering to a speech recognizer instead of a speach generator, because the Mac OS X already has a very good text-to-speech generator. (I once had a roommate that would have his computer convert everything to speech on his Mac)

[rickberry]

I believe this comment is nine years after the last comment on this thread. However, I have been using Dragon naturally speaking for more than that. Of time-probably 15 years. I recently tried to do indexing, and it does not appear that the fields Dragon fills are recognized as fields with data entered, even though it shows up on the screen. I'm having the same problem with just serve-if someone does happen to monitor this, perhaps it has to do with my chrome browser.

Unless there's a workaround or correction to a software glitch in just serve and family search, my ability to perform indexing is severely limited without using Dragon NaturallySpeaking.

[roblad]

Several years we tried using Dragon Dictation to help the hearing impaired to read a talk that was being given. This worked but at the time it was not good enough to use. Providing people every sunday who can do signing in ASL is a real challenge. It would be a wonderful contribution if it was possible to have a voice to text tool serve the hearing impaired. I saw a demo a couple years ago of a Japanese company who has a system that reads text and actually does signing using the image of a person. In the days of autonomous self driving vehicles, it ought to be possible to build a solution that could do this.

[drepouille]

I had a bright idea last week. I connected a ComTek receiver to the microphone jack of my laptop, and started a dictation web app to translate Sacrament Meeting to text. Unfortunately, I did not realize how weak my laptop battery was, and it died quickly. I will try again with a better battery next Sunday.

My greatest concern with using a web app is that our little DSL modem at the meetinghouse will not be fast enough to support speech to text.