Voice Banking

[rcharlton]

I'm not sure if I have limb onset or bulbar ALS - I had symptoms of both early on.

Complicating matters is the fact that I was born tongue tied - ie. my lingual frenulum is abnormally short and restricts mobility of the tongue tip - and had a speech impediment until I was 16 - I made the same sound for "th" and "s".

Anyway - I've always had a somewhat slurred speech - and it's been an effort to talk for some time - especially above noise or to the hearing impaired; however, I could always get the words out - until last Sunday. I was on the phone with a friend - and there were a couple of words that I could not enunciate or get out - my friend was subject to long silences on the phone while I worked on getting the words out. A pretty scary moment.

Coincident with this - two days earlier my fiance received an e-mail from a colleague suffering from ALS for over three years who now uses a voice box. One of her big regrets is that she did not record her own voice for use with the voice box before she lost the use of her speech.

I'm going to follow up on Al's referral to the Toront Rehabilition Clinic to see if they can give me some advice on voice banking.

Has anyone out there done any voice banking? What is the process like?

Thanks.

Richard

[Al]

Hi Richard. They gave me 4 options for voice banking. From free to about $50. Then you get about 14 pages of suggestions of things that you may want to bank. I haven't started yet as I am not having speech problems yet. Your OT should have no trouble getting you into TRC. I worked with Stephen Tse a very nice fellow, patient and kind. Good luck.

[rcharlton]

So I called the Toronto Rehab clinic about voice banking (thanks for the tip Al).

So there is no real magic to the process. It's merely a question of recording WAV files on your computer through a microphone. These sound files can then be used as in a speaking device. I was given a couple of questionnaires to direct my sound bites.

I also researched Voice Banking on the net and came across a program call Model Talker - which will actually synthesize an electronic voice based samples of your own voice.

I contacted the developer and here is what they told me:

"Thank you for your inquiry. Current development efforts on ModelTalker
are awaiting a grant funding decision; however, a functional alpha
version is still available from our research partner at
http://www.asel.udel.edu/speech/. Please let me know if I can assist you
in any way. There is a full tutorial included with the alpha system. We
ask that you begin by recording a few sentences and upload them so we
can look them over and make any suggestions / changes before you spend
the time to record the entire inventory. And we would always appreciate
any feedback you could provide us. We have a number of improvements
planned but user comments help us to better prioritize any new features
we would like to implement."

I've downloaded the program and let you know how it goes. The download was free.

There are a few key phrases that I want to make sure I record with my own voice - such as:

"Take a picture - it will last longer"
"I've got ALS - what's your excuse?" I should make this into a T-Shirt
"You're barking on thin ice" my father's favourite mixed metaphor
"Oops - sorry for the drool - I'm just happy to see you"
"I know you are but what am I?" the ultimate come back
"Do you want some fries with that shake"

Richard

[COlisa]

Hi RIchard,
Any luck with the voice banking?

My husband met with speech path today & she had never heard of it.

Did you get a usable result using ModelTalker?

Any info you can provide would be greatly appreciated!

Thanks,
Lisa

[rcharlton]

Hi Lisa,

I still have not finished completing the inventory.

INVtool is the actual voice banking software for model talker. It's amazing. You can add your own words and phrases. It gives you feedback on pitch, loudness and pronunciation - much better than recording through window recorder. Even if you don't plan to use modeltalker I'd recommend using INV tool to do voice banking.

It takes about 15 to 20 hours to complete the word inventory - and I only do 10 minutes at a time. I'm about halfway through the list.

I have to finish soon though cause my speech is deteriorating.

I'll keep you posted.

Richard

[MtPockets]

You said invatool is the software for the actual "model talker". I have no idea about any of this. What is a model talker? Is it like a typewriter? Big, small, heavy, light, works on your computer, portable, how long batteries last if it uses them.
Any clues about type of microphone to use to get realistic sound transfer? How much does all that cost?

Thanks for the info

God Bless
Capt. AL

[John1]

You said invatool is the software for the actual "model talker". I have no idea about any of this. What is a model talker? Is it like a typewriter? Big, small, heavy, light, works on your computer, portable, how long batteries last if it uses them....http://www.modeltalker.com/mt.php

[MtPockets]

Thanks John for the link.

Does anyone else use a different machine? Are there many different kinds.
Thanks

God Bless
Capt. AL

[Pearl]

This is a question that I contacted the speech augmentation rep at the Emory Clinic about for my friend. This is part of a reply I received and I thought it might be helpful to someone:


"There is a program called etriloquist we use all te time, that allows either saved phrases you just type in and use the computer voice, OR you can voice bank as you asked about. You have to of course record the phrases individually so it takes a while, but is very easy. Etriloquist comes with 14 folders with 227 pre-recorded phrases. These actually came from a survey of our GAALS patients so are appropriate.

NOTE: Make sure the patient hears the “AT&T VOICES” before they make a voice banking decision. For the 1st time ever (only happened 1 year ago) they digitized human voices. Over 95% of my patients (who would never use a regular computer voice) hear the AT&T voices and then see no reason to go to the trouble of recording their own – they are THAT good!"

[MarciaA]

My sister recorded over 400 words and phrases a couple weeks ago. She got a headset with an attached microphone and recorded them just on her computer at home. Here's a question I have and a concern Kathy (my sister) has...once she's using the computer to talk, how easy will it be to communicate? I'm sure it won't be as spontaneous as it is now, but how much of that will be lost?

Marcia

[rcharlton]

NOTE: Make sure the patient hears the “AT&T VOICES” before they make a voice banking decision. For the 1st time ever (only happened 1 year ago) they digitized human voices. Over 95% of my patients (who would never use a regular computer voice) hear the AT&T voices and then see no reason to go to the trouble of recording their own – they are THAT good!"

Thanks Pearl - but I can't ever imagine being happy about having someone else's voice.

Stephen Hamwking is British - and he says he hates having to speak with an American Accent through his machine.

The goal of Modeltalker is to synthesize speech based on your own voice - so you are not limited to an inventory.

The voice banking tool for Modeltaker is InvTool. Even if you don't plan on using Modeltalker I would URGE you to use InvTool to do your voice banking. You can add words to the inventory - it gives you feedback on pitch, loudness and pronunciation - it autosaves your recording and goes immediately to the next word on the list.

[rcharlton]

So I finally finished recording the inventory of phrases. It took me over a year - but I could devote only so much time every now and then. There was some degradation of my voice from when I started but it was still good enough.

There are about 1600 phrases and I would think you would need to dedicate a good 2 weeks of 1 or 2 hours a day.

But it is worth it. From those recorded phrases the Speech Research Laboratory
A.I. duPont Hospital for Children and the University of Delaware synthesized a voice based on my recorded phrases - and it sounds amazing. Far better than I expected. It still sounds a bit electronic - but it sounds better than any other voice synthesizer I've heard such as the one Stephen Heywood had - and best of all it sounds like ME!!

So now I can write anything (even if the words were not in my recorded vocabulary) - and it will be converted into speech using my voice.

I was almost in tears when I first heard it and I can't express what it means to know that, whatever happens to me, I will be able to communicate with my own voice.

The whole process and software are FREE and just requires some time commitment on your part.

I am forever indebted to everyone at the Speech Research Laboratory
A.I. duPont Hospital for Children and the University of Delaware

I would urge all PALS to do this now before it is too late.

http://www.asel.udel.edu/speech/ModelTalker.html

[CindyM]

Thanks everyone for some great advice. Rich, I'm especially interested in what else is on your list or pre-recorded phrases- other than the normal requests for food, assistance with daily tasks, and the like. I was LOL at your first ideas! Cindy

[MarciaA]

This is probably a stupid question, but now that you have everything recorded, how do you use it? Does a normal laptop suffice...do you need one with a touch screen...or do you need a different piece of technology altogether?

Marcia

[georgia]

We are also using Emory so we got the same idea. Since he is already losing speech to do it NOW but the site is down to download http://etriloquist.com/ I will look into the other.

[rcharlton]

I'm especially interested in what else is on your list or pre-recorded phrases- other than the normal requests for food, assistance with daily tasks, and the like.

Hi Cindy,

Many of the phrases are unconnected words - some are made up words - it is all about getting as many different sounds as possible.

With all of these sounds they can create a synthesized voice with an unlimited vocabulary - ANY word will be converted to your own synthetic speech.

[MtPockets]

I have recorded about 500 of the words and phrases so far. Since I am now having problems with the nerves at c2, c3, c4, I need to hurry up and finish. I am glad for you that you are finished. :-D I know how much trouble it is to keep repeating all those words until they are accepted.

Back to an earlier question, Can you use it on any computer, laptop, etc?
Can the whole thing be copied to another computer without much trouble?
Thanks for the info. I am planning on buying a laptop and just wanted to be sure I get one that will work with the model talker program.

God Bless
Capt AL

[Rhonda]

Just some info. for anyone using Emory in Atlanta GA. Steve Parker would be your contact for any of the speech programs or devices. He has been a great help to me.
Rhonda

[Montana Carol]

I've been posting about downloading ModelTalker and recording my phrases, etc. on a related thread -- just found your comments here, and am wondering if you're using ModelTalker now, and whether you're using it with a laptop or Dynavox or ?? Please give us an update, and thanks for your previous input. Carol

[rcharlton]

Fortunately - I can still speak on my own - so I'm not using Modeltalker yet - I did play with it on my PC with speakers - some phrases sound better than others - but I'm so relieved and happy to have it...

[Montana Carol]

I'm so glad to hear you're not having to use your ModelTalker voice yet, Richard; but I agree that it's such a relief to know it's there. I still haven't received my synthetic voice program back yet -- I think I'll e-mail and see if it's still in the works. Thanks for your reply. Carol

[Saubier]

I used Model Talker to bank my speech, but it was after several months of deterioration. It took me 6 hours to record all the phrases, but the result is good. People say it sounds just like me. I have bulbar onset, so I don't expect to have my voice for long, so I'm getting prepared. I think Tim, Jim, and James at Model Talker did a great job of "cleaning up" my voice, electronically.

Wouldn't it be great if you could get a voice from Johnny Cash or Kris Kristofferson?

Pappy.

[rmgraci]

I'm trying to figure out some voice banking solutions for my mother in law, and I'm a little overwhelmed by the options. Is ModelTalker the way to go? It seems like there's a lot of software that does this kind of thing. I gather that MT is the only one that uses words to string together sentences, and the others all just play pre-recorded entire phrases. Is that a correct analysis?

Once you have everything banked, which in itself is a task - how do you use it? What kind of hardware/software setups do you need to be equipped with to have the ModelTalker work for you?

Thanks!
- Ryan

[Saubier]

Ryan,
I got a Samsung Q1U-XP tablet PC because it's light and compact. Most portable PC speakers are not loud enough so I also got an auxiliary USB speaker.

Once you have ModelTalker loaded and set up you just input the text you want and when you hit Speak it comes out the speakers. You can use EZ Keys or SkeletonKey to accelerate your input, especially when you can't type anymore.

Pappy

[Montana Carol]

Pappy, do you find some phrases sound a lot better than others? I just received my finished synthetic voice and have it loaded into my Model Talker software. My recorded voice was pretty slow, and my s's weren't very good; so I guess that made some of my recordings not quite right, and some synthetic words also sound a little funny. But I'm still glad to have it, and it mostly sounds like me! I tried it out on a friend this morning, and she got tears in her eyes (in a good way!) to think somebody has made it possible for us to bank our voices. Carol

[rcharlton]

Pappy, do you find some phrases sound a lot better than others?

Some words sound great - others sound not as good.

Did you adjust the PCAT settings as they recommended?

Also remember that it's based on phonetics. So don't always stick with correct spelling.

For example - when I type in "bananas" it sounds unclear - but when I type in "baananas" - its sounds better.

Also - think about adding "." for pause. For example "Loralee" was unclear - but "Lora. Lee" was much better.

Good luck!

[Montana Carol]

Richard, thanks for your suggestions. I will try using phonetic spellings to help the sound on some of my synthetic sentences. I already figured out that I shouldn't use commas -- it causes a huge pause, longer than a period. I did do the suggested PCAT settings, and they told me to slow down the General Parameters/Speech Rate, too, to fit better with the slowness of my natural recording voice. I'm pretty happy with it now. Carol

[Saubier]

I did the ModelTalker voice banking, and it only took 6 hours (in 2 sessions). It's just software, but it's easy to use. I loaded it on a Samsung Q1U tablet because it's small and light to carry around, but also loaded it onto my PC at work and at home. I can still speak, slowly, but not for long, I think.

Pappy.

[John1]

I wish to start banking my voice with Model Talker. Can you suggest what type of microphone to use as I want to get the best quality recordings possible. MT recommends a head-mounted one but gives no further details. Here are some I'm looking at and would appreciate your input:
http://www.futureshop.ca/catalog/mfr.asp?man=Cyber+Acoustics&catid=2&logon=&langid=EN

[Thumbs Up]

I remember a time when my sister Roberta who loved shopping wanted me to take her to a mall. She wanted to treat me to lunch so we brought her lunch along, she was on a feeding tube. So we hooked her up for her lunch, a woman sitting near by kept staring and staring. So Roberta keyed into her communicater " What's wrong, haven't you ever seen someone eat before?"

[MtPockets]

Cyber Headset/ Microphone (AC-840) (http://www.futureshop.ca/catalog/proddetail.asp?logon=&langid=EN&sku_id=0665000FS10072933&catid=)

This is the one I use and there is only one main problem for me. The arm that the mike is on is too short to reach the front of my mouth, it ends about one inch before it. I also had a full beard when recording my model talker voice and the mike kept picking up noises of scraping my beard.

I have a very large head, hat size 8. Most people probably do not have this problem, but it would be nice if they listed the length of the boom piece. After a year of use the spring amount of the headset has decreased and it moves a lot. Falling off at times. I have no idea how to fix this problem.

Whichever you get, try to keep the mouth piece the same distance while recording.

God Bless
Capt AL

[John1]

Thanks for the input Al. I've noticed other complaints about this brand and have decided to go with one from Logitech. Also the $48 price at Future Shop is outrageous. I can find it online at many sites for <$20. The one I've ordered is:
http://www.tigerdirect.com/applications/SearchTools/item-details.asp?EdpNo=2484470&CatId=42

[Montana Carol]

Hi, All: While browsing the web, I came across the http://www.gusinc.com/ website which has the best speech devices, software and accessories I've found at very reasonable prices, approved by Medicare and MDA. One model uses the Samsung Q1U tablet PC mentioned here by Saubier awhile back, which is a regular computer weighing only 1.7 pounds. It comes with lots of stuff, and has options like a keyboard and scanning accessories for users who can't type. The synthetic voices that come with it are so great, if I get one I'll probably use that voice instead of my own. Pappy, are you liking your Q1U still?