View Full Version : Voice banking with Modeltalker

28th November 2014, 13:39
Has anyone got experience of recording their own voice on Predictable /Modeltalker ? Following Liam Dwyer s advice I want to record my own voice for use at a later date on some form of "text to speech" hardware. Clearly the loss of non verbal communication is inevitable but is it worth the considerable effort of speaking 1600 phrases ? Keep fighting it !

28th November 2014, 13:46
We had a demonstration from the people who provided the Predictable by Therapy application, which came through speech therapist. But sadly by that time my voice had gone. Get it done as soon as you can, Ian. You only need to record a few phrases, I only wanted "Hello darlings, how are you today" (for the grandchildren!) and "Can you get me a can of John Smiths Bitter" (for Peter!)
Ha! Ha! Sadly too late for me, but not for you. Much love Joycie x

28th November 2014, 17:38
As its too late for me to record my own voice I asked my daughter in the US to do it for me the other day. Her husband was going to set it up for her, probably over this come weekend. When I get any feedback I'll post here for you.


28th November 2014, 18:01
I tried to record my own voice using Modeltalker a few months back but could not even get beyond the initial calibration stage. I had put this down to the fact that my voice has already started to deteriorate but, not wanting to give in, I've just tried again with a slightly more expensive headset microphone and I've managed to record and submit the first ten phrases for review by the development team. I guess my message is use good equipment for the recording. If anyone wants to know which headset I've used just ask (don't want to give any free advertising:).

I should hear back in a few days as to whether my recordings are good enough to move on to the 1600 or so phrases required for full voice-banking so I'll give an update then.


28th November 2014, 18:53
Hi, in Germany we have a program called " Meine eigene Stimme "(My own voice). You have to read 7pages of syllables which was very demanding. A lady came to our house and recorded. Later when my voice is gone they will write a program I can use with tablet, PC, or other either with my hand or with my eyes. The name of the inventor is Dr. Mendel. He has got a HP where he says he is working on an English version. Arda

2nd December 2014, 15:21
Hi Artmart,
Just wondering how you got on and whether you intend to move on to the 1600 phrases? I'd like to know the make of headset, as we are in the process of setting up modeltalker voice banking for our Mum.

I will post how we get on for others to see, or feel free to pm me or ask on here.
Kind regards

2nd December 2014, 16:02
Hi Rachel

I heard back from Modeltalker support yesterday and my 10 test phrases were accepted so I've now been given acces to the full 1600 phrases and have started recording them today. It's not as bad as it sounds as some of the 1600 are single words but I've found that I needed to stop after the first 100 as the number of re-try's I needed to do were increasing! I have to work really hard with some pronunciation that used to need no effort or thought so the earlier your mum starts the better.

The headset microphone I found worked is a Sennheiser PC8 as below, I paid 34.99 on Amazon but there may be other suppliers who cost less. Set up with this was straightforward.

Good luck.



2nd December 2014, 18:30
Hi Rachel;

It is quite surprising how some cheap web cameras have quite good micro phones. I have been doing some recording for a push button talking device and the sound is quite realistic. So it might be worth a try if you have one.

Regards Terry

2nd December 2014, 19:14
Hi Terry

Modeltalker is quite demanding of a reasonable quality so they recommend a headset microphone to ensure consistency for each recording. The mic on a webcam, or anything used at a distance, will probably end up leading to problems setting up and using Modeltalker. I used what I thought was a reasonable headset mic by Logitech the first time I tried and it failed to even allow me to get past the initial calibration stage - if anyone wants to try any mic they might have to hand, that's the first bit to try.


2nd December 2014, 21:33
Hi Arthur;

That makes sense, probably a head mic is best as it will give you consistent noise levels, my recording could be altered to suit.

Thanks, Terry

4th December 2014, 00:03
Hi Arthur and Terry plus anyone interested in using Model Talker,
Thought you would like to see this advice from MT regarding using a mic v headset. We had problems with Mum's first calibration due to using an analogue headset. Model Talker seem to be keen to get it right for people and so far have a give me VERY quick and helpful advice. Amazing they are doing this for free!

Quote from MT......
"The important part of the headset is the boom-mounted microphone that keeps the mic very close to the speaker’s mouth and helps to hold exactly the same mouth-to-mic distance from one recording to the next. People who use a free standing mic must mount it on a boom to keep it very close to the speaker and use a pop screen between the mic and the speaker’s mouth. The speaker must also be used to recording this way and maintaining a very constant distance to the mic. For most people who are not professionals, the headset is just a much easier approach to manage.

As for recording with professional equipment, we would need to know more about the equipment that is available. Hopefully there would be a Windows PC that can run MTVR and be operable from within the sound booth. Often, the automatic calibration in MTVR will not work on professional equipment so we will need to go over how to do the manual calibration with someone who will be overseeing the equipment."

Best wishes

4th December 2014, 13:49
Hi Rachel;

It is so nice to get some help without bashing one's head against the wall. So many disability sites make it so hard, even if you are paying.

I got some great free advice off of Tony that runs the android app "free speech", he gave me links to other good free useful apps.

Regards to them and you, love Terry

PS:- Well done Paul

4th December 2014, 16:27
Hi All,
I have just passed the validation with ModelTalker (had to buy a new headset as my cheap Logitech one was not up to it). Ended up with a Sennheiser PC36 (amazon 32-98). You have to have the mic set to the side of your mouth and low down. Also the room need to be quiet and not too much echo off smooth walls (needs carpet, curtains etc).
I started yesterday on the 1600 and managed to get 100 done. It was quite hard work as I had to speak quite loudly and had to rework a few. The good thing is that you can start using "your voice" after just 25 contributions so I had fun trying in phrases and seeing how it sounded. I am quite impressed, but it will take a few more days dedicated hard work to get all 1600 done.
I will keep you updated about how i get along.

4th December 2014, 21:52
Hi Paul

Good to hear you're making progress and good points there about the room setting. I think it's worth sitting in the same room for all sessions of recording to ensure a standard to what goes into the whole set. I'm around 400 phrases in over 4 sessions, I stop when I start getting too many re-try's. I'm re-calibrating the mic for each session (I think Modeltalker recommend this) and, so far the recording are consistent. Good luck with progress.

Rachel - good to see you are getting useful support from MT. Hope it all goes well once your mum gets started.


7th December 2014, 13:10
Thanks guys for your feedback. I too have started the task of recording the 1600 phrases. It s not easy but, I find, 2 or 3 repeats of "failed" sentences do result in success. Tiredness is an issue as the tongue twisters do take some getting right. The more I try the more disjointed the longer words sound. A tip worth noting is re failures linked to pronunciation. I emailed MT on this point to be told that pronunciation is the least important to get right as they accept our English Yorkshire accents will inevitably sound odd to an American. So far so good. For those who are giving it a go keep going, it will be worth it ! Keep happy. Ian

7th December 2014, 15:19
Hi everyone currently working on recording phrases for Model Talker.
It's great to see the sharing of advice - I really appreciate it as we are having a testing time getting it set up. Tried again to get Mum's recordings calibrated on Saturday. After a lot of fiddling, because we couldn't get the system to hear the recordings, we submitted the test 10 phrases. Fingers crossed we will pass the test!
Best wishes all.

15th December 2014, 10:17
Hi - I was going well and got over 500 phrases done. PC has now developed a fault and is refusing to load. Just what I need at Christmas time! Another project to sort out. Better that siting idle in front of the tele I guess :-)

15th December 2014, 10:46
Is anyone using a Mac or are you pc set up? The reason I ask is that Santa is bringing me a Senheiser head Mike and my computer and tablet are Apple. I intend to start recording after Christmas.

15th December 2014, 12:52
Gd morning Barry, your ipad can use your synthesised voice with the Predictable app but you cannot use it to record. Only Windows 7 or better can be used to record. The MT email response service is excellent being 7 days a week. I suggest you use it to ask questions to sort out what you need. Once you are able to record it's easy and straight forward. The email address i have used is bunell@aseĺ.udel.edu for a chap called Tim. Good luck with it.

15th December 2014, 15:58
Hi Barry

If you get in touch with the folks at MT it might be worth asking if their Windows software will run in a Windows Emulator for the MAC. I've never used a MAC but a quick search of the Internet suggests there are a number of Windows Emulators out there (some free, some not) so this could be a way forward.

Good luck


19th December 2014, 10:21
Thanks for your replies guys. I intend to start recording in January and I will let you know how I get on.

24th December 2014, 12:50
Gd morning to you all. I completed the Model Talker voice banking last Friday and was surprised to receive the synthesised voice yesterday (Tuesday). I was a little shocked to hear my own voice but my wife and family all think it's good. I use a Samsung android tablet which is incompatible with Predictable & MT so I thought I would have to use my wife's Ipad. There is an app called Free Speech which costs nothing and is suitable for my tablet. So anyone can set the, selves up with their own voice and use a voice generating device Free Of Charge. That does make a Yorkshire man happy ! Have a good Christmas and keep fighting it ! Ian

24th December 2014, 14:41
Well done Ian;

Is the free speech app by Tony Atkins, he is very helpful if you have any problems. Unfortunately I can't get a free male voice on it and am stuck with a woman's one at the moment.


26th December 2014, 13:41
Thanks Terry. Yes it is By Tony Atkins. I could not get onto their website for instructions but an hour this morning has resulted in me being able to use it successfully. I quite like the posh ladies voice but I will use my own voice once I can work out how to download it. Keep happy !

26th December 2014, 14:39
I am stuck with the posh woman's voice Ian;

It is only a back up not that many ladies back me up now. Let us know how you get on and if there's any problems or things you don't like. You could email Tony if you're stuck or the Model talking people.


29th December 2014, 12:22
I am somewhat offended Terry that you prefer "a posh women's voice " to my manly Yorkshire voice but if that is what flicks your switch .....I have emailed Tony at Free Speech so I'm hoping he can solve my problem. I used Free Speech for an hour yesterday (no talking allowed) with my family and I was really pleased with how it went. The grandchildren really enjoyed the trial as I did. Keep happy to all in the MND club ! Ian

29th December 2014, 12:41
I am very sorry that I offended you Ian;

BUT I AM SURE YOU understand the implications of having an Yorkshire voice ! You were unfortunately enough to be bought up with it and didn't have any option but I have.

What device and operating system are you using.

Regards Terry

14th June 2015, 19:21
Terry just started the modeltalker filled in the form and sent off my test phrases. Just hope i havnt left it too late!. I hope it can recognise Essex,( around here a lot of people talk mockney I wonder just how putting on a cockney voice would work) just got to wait for the Email now. Colin.

bakeit Forum