Voice Recognition For VRS?
It’s widely agreed in the national Deaf Community, not just in Michigan’s, that there is a major dearth of interpreters who can voice accurately and smoothly. This issue is now spilling over to the VRS industry and one likely reason for this is because of the industry’s draining the pool of community/free-lance interpreters, who are notorious for their poor voicing skills.
I was at a meeting the other day, where this issue was raised and she wanted to know what actions she could take against the operator. This naturally sparked several side conversations. One of these side conversations involved me and Janet Jurus, who is the state interpreter and legislative lead of Michigan’s Division of Deaf and Hard of Hearing.
Janet proposed an idea that I think is excellent – have voice recognition software caption what the VRS operators are saying. That way, the majority of us would be able to track what the operator’s voicing, and be able to immediately correct the operator, if mistakes are made.
Since the majority of interpreters now work at least part-time for the VRS providers, it is my opinion that not only would this reduce problems, but also become an excellent component of the overall solution to ensure that interpreters make an earnest effort to keep their voicing skills up to par. It is easier to document the insufficient quality and ability of interpreters through the VRS, because you have the ability to record the calls, and with captions being part of the picture now, it would increase the validity of the Deaf Community’s complaints about the sub-par voicing skills of many interpreters.
12 Comments
Comments are closed.
kmayes said,
April 7, 2009 at 1:37 pm
I think it is a great idea. I have noticed that when I sign… VERY briefly… and the VRS would talk on and on, elaborating my signs and I wonder… what he/she is adding to my dialogues?
Good idea…
J.J. said,
April 7, 2009 at 2:17 pm
I am not sure voice recognition software is that GOOD yet.
Perhaps, if the same terp used the same software and we all downloaded info from that software…it might be okay. My understanding is that VR software works with the user and “self-corrects” over time depending on the user’s voice.
mishkazena said,
April 7, 2009 at 2:28 pm
What a terrific idea. The VRS companies do conduct their quality assurance programs, however they cannot check every relay call for quality assurance. We don’t have access to the data from these programs, either, so how do we know if they do rigorous quality checks or not?
More than few times, I’ve caught the interpreters misinterpreting me by lipreading.
This approach will help us monitor the quality of the VRS relay services ourselves as empowered consumers, especially for those who cannot lipread the interpreters.
Dianrez said,
April 7, 2009 at 2:58 pm
Excellent idea! It also has a double advantage: it could improve the word skills of the Deaf user and even help themselves correct their English or sign skills as well as that of the interpreter.
Too, we have regional differences where a sign can mean different things across the country. Hope the VRS industries pick up on this quickly.
White Ghost said,
April 7, 2009 at 3:21 pm
Ah. I remembered that several computer engineers are working on handshapes with the voice. It’s long way to go before they reach into the ASL and voices at the same time in the computer software.
Time will tell.
Just another opinion said,
April 7, 2009 at 3:39 pm
There’s WORK out there for CDI’s who can voice!
I don’t understand why the VRS companies aren’t SCREAMING for more CDI’s!
I for one, find that voicing is my forte, and I wish I could do more work in the VRS arena.
I don’t think that by using voice recognition software and VI’s it would be financially possible. The FCC wouldn’t fund both platforms.
Lucky
A Deaf Pundit said,
April 7, 2009 at 3:43 pm
All the VRS company would have to do is to install it in their programs. We already get the VP from several VRS companies, so it would be programmed in the VP. Then the interpreters/operators would use the same program.
It’s like making our phone numbers interoperable. I do think it is possible, and not financially prohibitive.
Lucky, do you mean CDIs who can actually use their voice? I think one reason CDIs aren’t widely used is because there’s not very many CDIs, and there’s just not enough knowledge about CDIs and what they do.
Valhallian said,
April 7, 2009 at 7:40 pm
isnt voice recognition already used with captel? if so, use that software or something similar eh? that would also mean that the VI would have to spend a few hours or so with that software so that it will recognize the VI’s voice. I would imagine that it would also have to be transferred on a server cuz how it works for one vi will not work for another vi when they switch unless they switch “voice databases” along with that as they switch places, but aye, wonderful idea.
Just another opinion said,
April 7, 2009 at 8:55 pm
Deaf Pundit,
Yes, I mean CDI’s who can use their voice. There are several of us, I know this for a fact.
You’re absolutely right, there’s not enough knowledge, AND there’s not enough acceptance of CDI’s as true interpreters. We’re still viewed by hearing interpreters as “consumers or clients.”
There is plenty of audism going around within the interpreting community!
We slip “outside” the traditional box of the CDI as defined by RID.
I use my voice. I lipread, I have a CI, and I’m fluent in ASL. The skills I have COULD be used by VRS providers if they’d just wake up!
Voicing is the most difficult thing for most interpreters, and we CDI’s who use our voices would be absolutely PERFECT for those times when there is difficulty voicing.
Lucky
Katie said,
April 7, 2009 at 9:58 pm
Voicing is the EASIEST part of interpreting, for me. It should be for most interpreters as well.
Most interpreters are NERDAs (Not Even Related to a Deaf Adult), which means ASL is their second language. In my experience, I always understand more than I can express, therefore voicing is much easier. HOWEVER, sometimes register choice can be a bit difficult to determine. Also, although it is easier for me than signing, it doesn’t mean that it’s good.
I just think that the Deaf community, and Deaf consumers, are familiar with the interpreting process and understand the hiccups associated with it more readily than hearing consumers who are not accustomed to the process of interpreting. Deaf consumers are more understanding of interpreter errors, which is why most interpreters think signing is easier. Voicing is riskier, because hearing consumers have very little patience for misunderstanding.
There are some awesome voice interpreters in Michigan. At the Deaf Health Fair at Madonna back in September, they had CART services captioning the interpreting. If you didn’t go, you should try to make it in a couple of years when they have it again. It was truly mind-blowingly awesome.
WAD said,
April 7, 2009 at 10:15 pm
Good suggestion but…
My biggest concern is the accuracy rate of the automatic speech recognition. Last time I heard it’s 95% and it’s still far from being reasonably acceptable to all people. This problem has been in the works for years and no breakthrough yet. I am waiting.
Robert Alfred Hawkins said,
April 8, 2009 at 5:45 am
This is an outstanding idea I thought of long time ago when I was involved with an innovated, start-up entity providing remote real-time transcription through Voice Recognition or Automatic Speech Recognition. This experience took place back in summer of 2002 through some point in 2003. That time the concept was very new compared to other models almost always requiring a number of on-site transcribers working onsite. There were imperfections, quirks especially from the get go. Overall, those who used were at least pleased. I’m not sure if today’s VR/ASR software are sophisticated enough to capture especially rapid back-and-forth video interpreting scenarios. Let’s see how this idea fly in the relay establishment. I’m sure there’ll be interpreters of all skills feeling threatened by this concept. Yes, some eminent interpreters just entertain their minds by freelancing whatever into their own spoken words.